- Email: [email protected]
A novel, element-based approach for the objective classification of bloodstain patterns
Forensic Science International 257 (2015) 220–228
Contents lists available at ScienceDirect
Forensic Science International journal homepage: www.elsevier.com/locate/forsciint
A novel, element-based approach for the objective classification of bloodstain patterns Ravishka M. Arthur a,*, Sarah L. Cockerton b, Karla G. de Bruin c, Michael C. Taylor d a
School of Chemical Sciences, The University of Auckland, Private Bag 92019, Auckland 1142, New Zealand Institute of Environmental Science and Research Ltd (ESR), Mount Albert Science Centre, Private Bag 92-021, Auckland 1142, New Zealand c Netherlands Forensic Institute, P.O. Box 24044, 2490 AA The Hague, Netherlands d Institute of Environmental Science and Research Ltd (ESR), Christchurch Science Centre, PO Box 29-181, Christchurch 8041, New Zealand b
A R T I C L E I N F O
A B S T R A C T
Article history: Received 19 May 2015 Received in revised form 25 August 2015 Accepted 31 August 2015 Available online 8 September 2015
The classification of bloodstain patterns has been identified as a challenging part of bloodstain pattern analysis due to the lack of a widely accepted and well-defined methodology and the ambiguity often associated with examining bloodstain patterns. The main aim of this study was to develop an objective, science-based method, for classifying bloodstain patterns, through the development of common language that could be used by BPA experts to describe the appearance of the pattern. This novel approach encourages a shift in the mindset of a BPA analyst by bringing them ‘back to the basics’ by treating components of a bloodstain pattern as discrete, observable and measurable units. One of the principal problems with current pattern classification methods is that pattern types are generally described in terms of the mechanism of pattern formation rather than grouping according to observable pattern characteristics. This study extends current BPA classification methodologies by developing and validating mechanism-free nomenclature that arises from observing and documenting the physical characteristics of bloodstain patterns. Following the grouping of bloodstain components on the basis of their physical characteristics, the formation evolution of these components is then investigated using concepts drawn from the fluid-dynamics of bloodstain pattern formation. This study offers a promising approach to distinguishing between different bloodstain pattern types through the use of visual aids in the form of colour maps, high-speed video and static digital images. ß 2015 Elsevier Ireland Ltd. All rights reserved.
Keywords: Bloodstain pattern analysis BPA Classification Forensic science Pattern
1. Introduction At the heart of bloodstain pattern analysis (BPA) lies the process of pattern classification. Ideally, bloodstain pattern classification is the product of the careful evaluation and identification of the characteristics of a pattern against objective and measurable classification criteria. Following this, pattern interpretation can occur in order to assist the crime scene reconstruction task. Despite the extensive dialogue in the BPA community around pattern classification, no one method has emerged that can claim to be fully validated and widely accepted. This is a result of a combination of problems that have created significant challenges for the analyst. One persistent problem is the tendency to describe patterns in terms of their proposed deposition mechanisms. This is evident in the current terminology list used by analysts e.g. [1],
* Corresponding author. Tel.: +64 2102954493. E-mail address: [email protected] (R.M. Arthur). http://dx.doi.org/10.1016/j.forsciint.2015.08.028 0379-0738/ß 2015 Elsevier Ireland Ltd. All rights reserved.
which is largely restricted to terms that describe deposition mechanisms, rather than pattern characteristics. This has the effect of pushing analysts to form conclusions about the cause of the pattern before a full analysis of the characteristics of the pattern. Another evident problem is the knowledge that without some additional key pattern characteristics, certain pattern types can be difficult to distinguish from one another. For example, a bloodstain pattern caused by exhalation of air as a result of injury to the mouth or airways can have similar characteristics to a pattern caused by blunt force trauma in the absence of vacuoles [2]. This ambiguity can lead to conflicting opinions by different analysts. Furthermore, a recent study has shown that BPA classification is vulnerable to biasing effects of contextual information [3]. While experienced BPA experts can reach complicated conclusions, they may experience difficulty in articulating the steps taken to arrive at these conclusions [4]. This problem has been explained by some, as a lack of a published methodology rather than the absence of a methodology per se. In addition to this, it has been suggested, that BPA training courses teach students the necessary
R.M. Arthur et al. / Forensic Science International 257 (2015) 220–228
221
skills involved in BPA (e.g. impact angle determination), but not how to apply these skills within an acceptable methodological framework [5]. The National Academy of Sciences (NAS) report [6] highlighted several issues hindering the quality of forensic science services in the United States of America. The report stated that forensic disciplines ‘‘need to develop rigorous protocols to guide these subjective interpretations and pursue equally rigorous research and evaluation programs’’ (NAS, 2009, p.8). Fingerprint identification, not unlike BPA, has been called out for insufficient validation of the current ACE-V method used for the analysis of fingerprint evidence [7]. It is fair to say that the classification of bloodstain patterns also remains largely a subjective interpretation and the uncertainties are large. The goal of this study was to encourage analysts to adopt a ‘‘back to basics’’ mind-set by asking them to treat bloodstain patterns as a combination of discrete, observable and potentially measurable units, which we termed ‘elements’. This approach was designed to avoid the use of mechanism-related jargon and to allow for the development of a common language composed of element descriptors that could be used objectively, to describe the characteristics of a pattern.
droplets were allowed to fall under gravity from the weapons, onto the rough side of a white cardboard sheet where they were left to dry overnight and imaged after 24 h. The different (weapon) surface types and the manipulation of drip height (30 and 100 cm) were conditions arbitrarily chosen to reproducibly create a variety of bloodstain elements. The second set of patterns (11 in total) was created using an inhouse impact device [9] designed to simulate an impact event. The device relies on the downward motion of an aluminium alloy rod impacting a pool of blood, propelled by the compression of restraining springs. For the patterns generated in this way, a pool of about 5 ml of blood was used and some variation was achieved by varying the degree of spring compression, in a qualitative fashion. Bloodstains were similarly collected on the rough side of white cardboard sheets and left to dry prior to imaging. A Photron FASTCAM SA1 HSDV high-speed video camera was used during bloodstain pattern creation to capture the evolution of the pattern formation. Following drying, digital still imaging with a Nikon D7000 Camera and Nikkor 60 mm macro lens was used to capture the resulting static pattern. The image resolution was 35 pixels/mm measured by ImageJ 1.48 v software (Public domain, http://imagej.nih.gov/ij; accessed August 2015).
2. Experimental methods
2.3. Deriving preliminary descriptions of elements
2.1. Overview
The laboratory-generated patterns were used to derive preliminary descriptions of a number of elements. Patterns were initially divided into a grid composed of 10 cm by 10 cm squares. Each square of the grid was methodically inspected for discrete units, which could be uniquely defined as an element. Each element within a square was documented in terms of its general appearance, size, colour and nature of its margins using simple shape-based descriptors.
At the outset of this study, a number of elements were defined, through visual assessment of some laboratory-prepared bloodstain patterns. An element was defined as a single bloodstain consisting of uninterrupted margins. Therefore, a bloodstain pattern could be considered to comprise of a number of individual elements. Each element was further distinguished on the basis of measurable properties (e.g. length and width in the case of elliptical stains) or a characteristic appearance (e.g. colour, shape). These definitions used simple, plain language and where applicable, shape-based terminology. Following this, element descriptions incorporating this terminology were collated to form a preliminary classification scheme. This scheme was primarily based on elements grouped on the basis of their appearance. High-speed video was used to understand the element formation evolution and to assist with distinguishing any observable and measurable element characteristics. The scheme was later refined using a layparticipant workshop and resulted in the creation of an ‘Atlas of Elements.’ The Atlas was tested and validated via a second, expertparticipant workshop. Finally, possible associations between elements were assessed using colour maps to visualise the element groups in a pattern. It was hoped that such associations might lead to a better understanding of the overall pattern and form the basis of an objective pattern identification. 2.2. Bloodstain pattern creation Two sets of bloodstain patterns were created in the laboratory using two different mechanisms, chosen to represent two common bloodstain patterns found at crime scenes. Porcine blood (an accepted blood substitute [8]) was used to generate these patterns. This blood was treated with aqueous acid citrate-dextrose (ACD) anticoagulant and kept refrigerated at all times prior to use (warmed up to 378C when required and blood used within 20 days from collection). Three replicate patterns were produced for each set. The first set of patterns (four in total) was created using an automated NE-300 Just InfusionTM syringe pump that pushed blood through plastic tubing, across the surface of two different ‘weapons’ (a cylindrical baton and a knife). Two or more blood
2.4. The classification scheme Element descriptors were combined to form a whole description, which collectively described a group of elements, which shared visible characteristics pertaining to their shape, margins and approximate dimensions. Each group was allocated a unique alphanumeric code, which was associated with observable element characteristics. This process ultimately led to the construction of a preliminary classification scheme. This classification scheme was further refined with the aid of a lay-participant workshop. The specific objectives for the workshop were (a) to test whether elements could be described in a consistent manner and without reference to a bloodstain pattern formation mechanism (b) to generate new descriptors for defining elements and (c) to determine whether element descriptions could then be used to reliably classify elements into meaningful groups. A total of eight individuals, primarily postgraduate students within the Forensic Science programme at The University of Auckland, New Zealand formed the lay-participant cohort. These participants had a science background, minimal BPA knowledge and no practical forensic experience. The group was chosen to provide a fresh look at terminology, by minimising the discipline jargon and potential biasing effect of existing methodologies. Ethics approval for the use of human participants was obtained from the University of Auckland Human Participants Ethics Committee, New Zealand (Ref: 8037). The workshop comprised five separate activities, which were conducted and completed simultaneously by all participants and with time restrictions imposed (1–2 min per task). A total of seven different element images were chosen for the various workshop tasks and were selected on the basis that they
R.M. Arthur et al. / Forensic Science International 257 (2015) 220–228
222
were an accurate representation of the element group description in question. The workshop activities were as follows: Activity 1: Participants were asked to draw sketches to reflect single element descriptors, taken from the preliminary classification scheme. Activity 2: Participants were asked to draw sketches in response to a more complete description of an element group. Activity 3: Participants were provided with a list of element descriptors selected from the preliminary classification scheme together with an image of an element. They were then asked to select those terms that they considered best described the appearance of the element. Activity 4: Participants were provided with images of three different element types. They were asked to describe the appearance of the element using common language descriptive terms of their own choosing. Activity 5: Images of different elements were provided together with the preliminary classification scheme. Participants were asked to use the scheme to select the most appropriate element group that best described the appearance of the element. Using responses from the workshop, the preliminary classification scheme was reviewed and modified and a visual aid termed ‘The Atlas of Elements’ was devised. This document consisted of modified element group descriptions and a combination of high speed and static images to characterise the element group. The role of the fluid dynamic principles of bloodstain formation was considered whilst studying the formation of the individual elements. In general, elements in a group were observed to follow a similar formation evolution. Understanding the fluid dynamics assisted with the identification of any observable and measurable element characteristics. This was accomplished by studying highspeed images of element formation.
The Atlas of Elements and a Glossary of Terms was provided to the BPA-trained participants, together with 17 images (at least two replicates per element group) representing the various elements. Participants were asked to classify the elements based on the groups designated by the Atlas. A successful classification was deemed to have occurred if the element was placed in the intended element group. 2.6. Characterising associations The digital images of the 15 laboratory-prepared bloodstain patterns (11 impact patterns, four-drip patterns) generated at the start of the study were used for the characterisation of associations. These digital images were converted into colour maps using Adobe Photoshop CS5, by inspecting each individual element in a single pattern and classifying it using the descriptions within the Atlas of Elements. A different colour was allocated for each element group and a coloured circle placed in the approximate location in the pattern for each element. Colour maps were created to visualise all element groups and to determine whether any associations could be identified for a particular pattern. Associations were characterised according to their spatial location and considered as either: Intra-element: The relationship within an element group, for example, between elements in Group 1A or elements in Group 1B or Inter-element: The relationship between element groups, for example, between Groups 1A and 3A. By creating colour maps, it was hypothesised that relationships within and between element groups could be visually assessed and their contribution to the distinctives of the overall pattern could be determined. 3. Results and discussion 3.1. Workshop
2.5. Validation of the atlas of elements A classification test was designed to test the use of the Atlas of Elements. This test was performed by an expert-participant group. The expert-participant group consisted of 17 participants. A total of 12 employed by ESR Ltd, were selected because they were BPA qualified and had the authority to carry out BPA reporting in casework. The remaining five participants, employed by the Netherlands Forensic Institute (NFI) held equivalent reporting qualifications.
The lay-participant workshop, composed of five different activities was used to validate and refine the descriptors that were generated to describe the physical appearance of elements. A complete analysis of the results of this workshop is reported in [10]. Fig. 1 shows the results of Activity 1, where participants drew sketches to reflect single element descriptors. The similarities in the drawn images suggest that a common understanding of well-defined element descriptors can be
Fig. 1. Participant drawings in response to the descriptor ‘smooth continuous boundary’.
R.M. Arthur et al. / Forensic Science International 257 (2015) 220–228
223
Fig. 2. Participant drawings in response to the description ‘oval base which tapers inwards to form a slim neck’.
achieved. It was evident that most participants tended to associate the terms ‘smooth continuous boundary’ with a circle shape. Fig. 2 shows the results of Activity 2, where participants drew sketches in response to the element description ‘an oval base which tapers inwards to form a slim neck.’ The sketches demonstrate a consistent understanding of the complete element description as well as the ability to recognise that elements can have different parts (i.e. a base and a neck have been drawn in different proportions). It is clear that element descriptions could be enhanced in the future by the addition of quantitative parameters. For example, the size of the neck and base could be expressed as a ratio so that this element could be better defined. Fig. 3 shows an example of the results of Activity 3 where a combination of terms from the preliminary classification system were selected by participants as best descriptors for the element in question. Simple shape-based terms, such as ‘circular’ and ‘uneven edges’ were used by all participants to describe the element shown in this example. This result provided the basis for incorporating these particular terms into the final element descriptions within the Atlas of Elements. Other terms, although less popular were selected as alternative descriptors (‘finger-like projections’ and ‘circular/oval droplet’) for this
element also. Activity 3 was most important for filtering out terms that could be misused. For example, the term ‘finger-like’ was used by 3 participants, however, upon review, this term was eliminated from the final classification system because it compared the appearance of the element to a physical object and therefore inherited the variability of the shape of that object. One of the main objectives of the workshop was to generate new descriptors for defining elements by giving participants an opportunity to describe the element using simple terms. This is illustrated in Fig. 4 which demonstrates the results of Activity 4, where participants described images using descriptive language of their own choosing. This activity gave participants more flexibility in the choice of terms and the opportunity to describe different parts of the element. Overall, a wide range of descriptive terms were generated for the elements provided. Participants generally had the tendency to describe elements in terms of regions. For example, as shown in Fig. 4, some terms clearly corresponded to the central part of the element (labelled Region 1) and others corresponded to the narrow part of the element (labelled Region 2). The results also showed that terms with similar meanings, such as ‘tail’ or ‘extension’ need to be well-defined if they were to be used to describe different features.
Fig. 3. Image of element and frequency of terms used by participants to describe it.
224
R.M. Arthur et al. / Forensic Science International 257 (2015) 220–228
Fig. 4. Examples of the results of Activity 4 showing two image of elements and the range of terms produced by participants. Terms are listed according to the main regions of the element.
The final objective of the workshop was to conduct a classification task in order to determine whether element descriptions could be used to reliably classify elements into meaningful groups. Examples of the results of Activity 5 are shown in Fig. 5.
For all three of the element examples shown in Fig. 5, 75% or more of the participants selected the group description intended to be associated with that element demonstrating that a high degree of classification precision is possible.
Fig. 5. Images of groups 3A, 3B and 4 elements provided to participants and their respective classification frequencies.
R.M. Arthur et al. / Forensic Science International 257 (2015) 220–228
225
Table 1 Excerpt from the Atlas of elements: Group 3B elements. Group 3B elements
Fluid dynamic mechanisms
High-speed video stills
Element attributes shape: Oval with a single projection referred to as a tail. Projection consists of a rounded tip and can appear partially transparent. Projection curves inwards on both sides between the tip and the base of the oval to form a slim neck.
1. Upon impact of a free-falling droplet onto the surface of a preexisting blood film, a sheet of blood (forming a distinct crown shape) forms due to a transfer of kinetic energy into surface energy (i.e. radial expansion).
2. Group 3B elements begin their early stages of formation from a secondary volume of blood that has detached from the original sheet. Sheet break-up and detachment occurs due to the growth of capillary waves travelling along the sheet.
Diagrammatic example
3. A droplet and its associated ligament detach from this secondary volume due to viscous shear and the surface tension force. 4. The surface tension force causes prolate and oblate oscillations which eventually die away leaving the blood in a spherical shaped droplet. Retraction of the ligament is seen as a result of the surface tension force minimising the surface area. 5. As the droplet impacts the surface at a low impact angle and with low energy, a transfer of kinetic to surface energy will drive spreading along the surface. A larger volume of blood is found at the base which causes it to appear raised. 6. Spreading is propagated along the surface. The surface tension force can be seen to act on the spherical tip.
Examples of variation
7. Retraction as a result of the surface tension force pulls some volume of blood to the centre of the element. The early formation of the projection is seen during this stage. This projection is limited in length because of the low energy of the impacting droplet. 8. Surface tension and viscous shear eventually slow down and halt the spreading motion. The single projection (tail) is more pronounced as spreading has eventually halted. The blood retains a raised profile. 9. Upon drying, the resulting element is an oval shape with a single projection referred to as a tail. Grid size: 10 mm 10 mm
3.2. Atlas of elements The classification scheme was further enhanced by a study of high-speed video images of examples of the formation of elements from each group. These images, along with static images of typical examples of resulting elements, were combined to form an ‘Atlas of Elements.’ An excerpt from this Atlas is shown in Table 1. The full Atlas in its present stage of development is presented in [10]. 3.3. Validation test of the atlas of elements The results of the validation test of the Atlas are shown in Fig. 6. Blue bars correspond to replicate one, that is the first example of an element in each group and the red bars correspond to replicate two. Fig. 6 indicates a high degree of precision was evident when the Atlas was used by BPA experts to classify elements. An overall mean classification success rate of 86% was achieved. All element examples in Groups 1B and 4 were successfully classified by all participants. Of interest, was the low success rate for one example of the elements in groups 1C, 2A and 3A. This suggests that some
elements are more variable in appearance than others and hence, more susceptible to misclassification. It is likely that these elements are also more difficult to describe unambiguously. To improve the classification rate, a larger database of example images may help analysts better understand the differences within these groups and a more in-depth study of their formation mechanism may shed light on these differences. Additionally, developing descriptions that better capture the essential distinctives of these groups might also help. Finally, changing the overall number of elements could have a bearing on the classification success rate 3.4. Characterising associations A total of 15 laboratory-generated bloodstain patterns were analysed for element associations. Associations were defined as relationships between elements in one group (intra-element) and between elements in different groups (inter-element). It was hypothesised that these associations could be used to distinguish between different bloodstain pattern types. The colour maps outlined in Figs. 7 and 8 document two examples of laboratorycreated patterns.
226
R.M. Arthur et al. / Forensic Science International 257 (2015) 220–228
Fig. 6. Classification results of elements by 17 BPA experts (*15 experts only) when using the Atlas of Elements to classify examples of the different elements, R1 and R2 refer to replicates 1 and 2, respectively.
Fig. 7 demonstrates a tightly clustered intra-element association between Group 1A (blue) elements in the lower central region of the pattern and a more dispersed grouping towards the periphery. These characteristics were reproducibly found in those patterns created with the in-house impact device. A scattered association between elements in Group 3B (black) also occurred in the impact patterns and was commonly found at the periphery of the pattern. Finally, all the impact patterns analysed in this study provided evidence of a scattered, mid-pattern association between Group 1B (pink) elements with a few isolated elements at the periphery. This mid-pattern association was also true for Group 3A (green) elements. There is a possibility that analysis of further patterns may reveal a consistent inter-element association between Groups 1B (pink) and 3A (green) elements for such patterns. Patterns generated as a result of dripping blood, tended to exhibit fewer elements dominated by Group 1A (blue) elements,
Fig. 7. Digital image of bloodstain pattern (left) and corresponding colour map (right). AC (area of convergence) has been specified. Groups; 1A (blue), 1B (pink), 1C (purple) 3A (green), 3B (black), 5 (red), 6 (yellow).
Fig. 8. Digital image of bloodstain pattern (left) and corresponding colour map (right). Group 1A (blue) Group 3A (green) around single group 2B (brown) element.
Fig. 9. Digital image of bloodstain pattern (left) on rear bumper of vehicle and corresponding colour map (right).
R.M. Arthur et al. / Forensic Science International 257 (2015) 220–228
which were typically randomly distributed within the pattern (see Fig. 8). To illustrate the possible application of this approach to casework, Fig. 9 exhibits an image of the rear of a vehicle with confirmed bloodstaining. A colour map of the elements identified in this pattern is also shown (Fig. 9). The analysis indicated that Group 1A (blue) elements, although fewer in number, were densely clustered near the centre of the pattern. A few Group 3B (black) elements were located at the periphery of the pattern and a mid-pattern scattering of Group 1B (pink) elements was noted. The apparent inter-element association between Groups 1B (pink) and 3A (green) elements was also noted for some closely spaced element pairs. Overall, the pattern exhibited some of the characteristic element associations noted for laboratory-created impact spatter patterns. 4. Conclusions Ultimately, BPA analysts aim to achieve two things. First, to reliably determine the likely mechanism(s) of formation of a bloodstain pattern and secondly, to be able to unambiguously articulate their conclusions and demonstrate how they were reached. To aid the analyst in these objectives, a novel, more objective classification methodology has been explored. The study provides a fresh look at the basics of bloodstain pattern analysis by focussing on the description of individual pattern components, termed ‘elements’ and analysing their associations within the pattern. In doing so, this method aimed to address some of the problems that have affected current classification methodologies. The results of this study show that, not only is it possible to break down a bloodstain pattern into components (elements), but it is possible to also describe these components using nonmechanism related nomenclature. Elements were grouped according to their appearance and how they were created, aided by the study of high-speed video images of element formation. Results indicated that it was possible to consistently describe elements and to distinguish between different element groups. There was also a tendency for certain groups of elements to be found in one specific bloodstain pattern. However, the validity of these results needs to be investigated further. In addition to this, it became apparent that a consideration of associations within and between element groups could form the basis for distinguishing different types of bloodstain patterns. The methodology presented in this study offers several potential advantages over current BPA methodologies. Firstly, the development of a common mechanism-free language for the sole purpose of describing the attributes of a pattern appears to be realistic. Secondly, a visual aid in the form of a prototype Atlas of Elements, that supports the analyst in his/her classification decision-making process, has been produced. The Atlas of Elements showcases a visual summary of the element-based classification system that was developed within this study. It is potentially a powerful tool that, with further development, could be used and easily understood by any expert to aid in the pattern classification process. The user can refer to a validated description that has incorporated simple mechanism-free terminology, static images comprised of examples of the element and finally a summary description of the fluid dynamics of element formation. The latter is supported by a series of in-depth high-speed video images, which documents significant detail of the formation stages of the element. It is evident that the method proposed within the Atlas of Elements not only offers the discipline a novel approach to classification but also capitalises on the use of imagery to support and strengthen expert interpretation. This should have the added benefit of making this an easy-to-articulate process in a court setting. Thus, bloodstain pattern classification could become a
227
demonstrably more objective process. Furthermore, the Atlas offers promise as a reference document that, if widely accepted, could bring added consistency to BPA. Finally the classification method proposed here is strongly rooted in the fundamental science of fluid dynamics. It should be noted that the concept proposed here remains a work in progress. Many more patterns need to be studied and the use of the classification method rigorously tested before its use in casework could be contemplated. The vital link between associated elements in a pattern and the mechanism of pattern formation needs to be established and validated. Furthermore, this study was limited to bloodstains on cardboard surfaces and only two of the major bloodstain pattern formation mechanisms (impact and drip). It is accepted that with changes in the target substrate and consideration of other pattern mechanisms, there will be other element types and associations to consider. This will further enrich the classification methodology proposed. Although the stated objectives of this study included the development of a more objective methodology, it should be clear that the classification methodology proposed still contains a subjective component. However, the proposed method lends itself to the addition of computer-vision based tools. Automated methods are already used in several forensic applications, such as, the comparison of striation marks resulting from fired bullets [11] and toolmark impressions [12] as well as for shoeprint retrieval and matching [13]. Statistical models such as principal component analysis have also been used in the study of accidental marks found on footwear [14,15]. It may also be possible to derive quantitative information about a bloodstain pattern using similar methods and use this information to develop a more automated and robust process for an element-based bloodstain pattern classification method. Acknowledgements The authors would like to acknowledge Dr. Douglas Elliot (ESR and School of Chemical Sciences, University of Auckland), Dr. John Buckleton (ESR) and Dr. Mark Jermy (University of Canterbury). Thank you to the participants from ESR and NFI for taking part in the experiments. This work was carried out in fulfilment of the Master of Science (Forensic Science) programme through the University of Auckland, New Zealand. The financial support of ESR through the ESR Masters Scholarship is sincerely appreciated. References [1] SWGSTAIN, Scientific working group on bloodstain pattern analysis: recommended terminology, Forensic Sci. Commun. 11 (2009). [2] T. Bevel, R.M. Gardner, Bloodstain Pattern Analysis with an Introduction to Crime Scene Reconstruction, CRC Press, Boca Raton: FL, USA, 2008. [3] T. Laber, P. Kish, M. Taylor, G. Owens, N. Osborne, J. Curran, Reliability Assessment of Current Methods in Bloodstain Pattern Analysis, National Institute of Justice, U.S. Department of Justice, 2014 Report No: 2010-DN-BX-K213. [4] J. Saviano, Articulating a concise scientific methodology for bloodstain pattern analysis, J. Forensic Identif. 55 (4) (2005) 461–470. [5] R. Gardner, Defining a methodology for bloodstain pattern analysis, J. Forensic Identif. 56 (4) (2006) 549–557. [6] Committee on Identifying the Needs of the Forensic Sciences Community, National Research Council, Strengthening Forensic Science in the United States: a Path Forward, National Academies Press, 2009 February. [7] L. Haber, R.N. Haber, Scientific validation of fingerprint evidence under Daubert, Law Prob. Ris. 7 (2008) 87–109. [8] M.A. Raymond, E.R. Smith, J. Liesegang, The physical properties of blood-forensic considerations, Sci. Justice 36 (1996) 153–160. [9] T. Stotesbury, The development of sensitivity tests for the creation and use of synthetic blood substitutes in bloodstain pattern analysis, in: Unpublished master’s thesis, The University of Auckland, New Zealand, 2012. [10] R. Arthur, The development of an element-based method for the reliable identification of bloodstain patterns, (Unpublished master’s thesis), The University of Auckland, New Zealand, 2013.
228
R.M. Arthur et al. / Forensic Science International 257 (2015) 220–228
[11] F.P. Leo´n, Automated comparison of firearm bullets, Forensic Sci. Int. 156 (2006) 40–50. [12] M. Baiker, I. Keereweer, R. Pieterman, E. Vermeij, J. van der Weerd, P. Zoon, Quantitative comparison of striated toolmarks, Forensic Sci. Int. 242 (2014) 186–199. [13] G. AlGarni, M. Hamiane, A novel technique for automatic shoeprint image retrieval, Forensic Sci. Int. 181 (2008) 10–14.
[14] N.D.K. Petraco, C. Gambino, T.A. Kubic, D. Olivio, N. Petraco, Statistical discrimination of footwear: a method for the comparison of accidentals on shoe outsoles inspired by facial recognition techniques, J. Forensic Sci. 55 (1) (2010) 34–41. [15] D. Sheets, S. Gross, G. Langenburg, P. Bush, M. Bush, Shape measurement tools in footwear analysis: a statistical investigation of accidental characteristics over time, Forensic Sci. Int. 232 (2013) 84–91.
Contents lists available at ScienceDirect
Forensic Science International journal homepage: www.elsevier.com/locate/forsciint
A novel, element-based approach for the objective classification of bloodstain patterns Ravishka M. Arthur a,*, Sarah L. Cockerton b, Karla G. de Bruin c, Michael C. Taylor d a
School of Chemical Sciences, The University of Auckland, Private Bag 92019, Auckland 1142, New Zealand Institute of Environmental Science and Research Ltd (ESR), Mount Albert Science Centre, Private Bag 92-021, Auckland 1142, New Zealand c Netherlands Forensic Institute, P.O. Box 24044, 2490 AA The Hague, Netherlands d Institute of Environmental Science and Research Ltd (ESR), Christchurch Science Centre, PO Box 29-181, Christchurch 8041, New Zealand b
A R T I C L E I N F O
A B S T R A C T
Article history: Received 19 May 2015 Received in revised form 25 August 2015 Accepted 31 August 2015 Available online 8 September 2015
The classification of bloodstain patterns has been identified as a challenging part of bloodstain pattern analysis due to the lack of a widely accepted and well-defined methodology and the ambiguity often associated with examining bloodstain patterns. The main aim of this study was to develop an objective, science-based method, for classifying bloodstain patterns, through the development of common language that could be used by BPA experts to describe the appearance of the pattern. This novel approach encourages a shift in the mindset of a BPA analyst by bringing them ‘back to the basics’ by treating components of a bloodstain pattern as discrete, observable and measurable units. One of the principal problems with current pattern classification methods is that pattern types are generally described in terms of the mechanism of pattern formation rather than grouping according to observable pattern characteristics. This study extends current BPA classification methodologies by developing and validating mechanism-free nomenclature that arises from observing and documenting the physical characteristics of bloodstain patterns. Following the grouping of bloodstain components on the basis of their physical characteristics, the formation evolution of these components is then investigated using concepts drawn from the fluid-dynamics of bloodstain pattern formation. This study offers a promising approach to distinguishing between different bloodstain pattern types through the use of visual aids in the form of colour maps, high-speed video and static digital images. ß 2015 Elsevier Ireland Ltd. All rights reserved.
Keywords: Bloodstain pattern analysis BPA Classification Forensic science Pattern
1. Introduction At the heart of bloodstain pattern analysis (BPA) lies the process of pattern classification. Ideally, bloodstain pattern classification is the product of the careful evaluation and identification of the characteristics of a pattern against objective and measurable classification criteria. Following this, pattern interpretation can occur in order to assist the crime scene reconstruction task. Despite the extensive dialogue in the BPA community around pattern classification, no one method has emerged that can claim to be fully validated and widely accepted. This is a result of a combination of problems that have created significant challenges for the analyst. One persistent problem is the tendency to describe patterns in terms of their proposed deposition mechanisms. This is evident in the current terminology list used by analysts e.g. [1],
* Corresponding author. Tel.: +64 2102954493. E-mail address: [email protected] (R.M. Arthur). http://dx.doi.org/10.1016/j.forsciint.2015.08.028 0379-0738/ß 2015 Elsevier Ireland Ltd. All rights reserved.
which is largely restricted to terms that describe deposition mechanisms, rather than pattern characteristics. This has the effect of pushing analysts to form conclusions about the cause of the pattern before a full analysis of the characteristics of the pattern. Another evident problem is the knowledge that without some additional key pattern characteristics, certain pattern types can be difficult to distinguish from one another. For example, a bloodstain pattern caused by exhalation of air as a result of injury to the mouth or airways can have similar characteristics to a pattern caused by blunt force trauma in the absence of vacuoles [2]. This ambiguity can lead to conflicting opinions by different analysts. Furthermore, a recent study has shown that BPA classification is vulnerable to biasing effects of contextual information [3]. While experienced BPA experts can reach complicated conclusions, they may experience difficulty in articulating the steps taken to arrive at these conclusions [4]. This problem has been explained by some, as a lack of a published methodology rather than the absence of a methodology per se. In addition to this, it has been suggested, that BPA training courses teach students the necessary
R.M. Arthur et al. / Forensic Science International 257 (2015) 220–228
221
skills involved in BPA (e.g. impact angle determination), but not how to apply these skills within an acceptable methodological framework [5]. The National Academy of Sciences (NAS) report [6] highlighted several issues hindering the quality of forensic science services in the United States of America. The report stated that forensic disciplines ‘‘need to develop rigorous protocols to guide these subjective interpretations and pursue equally rigorous research and evaluation programs’’ (NAS, 2009, p.8). Fingerprint identification, not unlike BPA, has been called out for insufficient validation of the current ACE-V method used for the analysis of fingerprint evidence [7]. It is fair to say that the classification of bloodstain patterns also remains largely a subjective interpretation and the uncertainties are large. The goal of this study was to encourage analysts to adopt a ‘‘back to basics’’ mind-set by asking them to treat bloodstain patterns as a combination of discrete, observable and potentially measurable units, which we termed ‘elements’. This approach was designed to avoid the use of mechanism-related jargon and to allow for the development of a common language composed of element descriptors that could be used objectively, to describe the characteristics of a pattern.
droplets were allowed to fall under gravity from the weapons, onto the rough side of a white cardboard sheet where they were left to dry overnight and imaged after 24 h. The different (weapon) surface types and the manipulation of drip height (30 and 100 cm) were conditions arbitrarily chosen to reproducibly create a variety of bloodstain elements. The second set of patterns (11 in total) was created using an inhouse impact device [9] designed to simulate an impact event. The device relies on the downward motion of an aluminium alloy rod impacting a pool of blood, propelled by the compression of restraining springs. For the patterns generated in this way, a pool of about 5 ml of blood was used and some variation was achieved by varying the degree of spring compression, in a qualitative fashion. Bloodstains were similarly collected on the rough side of white cardboard sheets and left to dry prior to imaging. A Photron FASTCAM SA1 HSDV high-speed video camera was used during bloodstain pattern creation to capture the evolution of the pattern formation. Following drying, digital still imaging with a Nikon D7000 Camera and Nikkor 60 mm macro lens was used to capture the resulting static pattern. The image resolution was 35 pixels/mm measured by ImageJ 1.48 v software (Public domain, http://imagej.nih.gov/ij; accessed August 2015).
2. Experimental methods
2.3. Deriving preliminary descriptions of elements
2.1. Overview
The laboratory-generated patterns were used to derive preliminary descriptions of a number of elements. Patterns were initially divided into a grid composed of 10 cm by 10 cm squares. Each square of the grid was methodically inspected for discrete units, which could be uniquely defined as an element. Each element within a square was documented in terms of its general appearance, size, colour and nature of its margins using simple shape-based descriptors.
At the outset of this study, a number of elements were defined, through visual assessment of some laboratory-prepared bloodstain patterns. An element was defined as a single bloodstain consisting of uninterrupted margins. Therefore, a bloodstain pattern could be considered to comprise of a number of individual elements. Each element was further distinguished on the basis of measurable properties (e.g. length and width in the case of elliptical stains) or a characteristic appearance (e.g. colour, shape). These definitions used simple, plain language and where applicable, shape-based terminology. Following this, element descriptions incorporating this terminology were collated to form a preliminary classification scheme. This scheme was primarily based on elements grouped on the basis of their appearance. High-speed video was used to understand the element formation evolution and to assist with distinguishing any observable and measurable element characteristics. The scheme was later refined using a layparticipant workshop and resulted in the creation of an ‘Atlas of Elements.’ The Atlas was tested and validated via a second, expertparticipant workshop. Finally, possible associations between elements were assessed using colour maps to visualise the element groups in a pattern. It was hoped that such associations might lead to a better understanding of the overall pattern and form the basis of an objective pattern identification. 2.2. Bloodstain pattern creation Two sets of bloodstain patterns were created in the laboratory using two different mechanisms, chosen to represent two common bloodstain patterns found at crime scenes. Porcine blood (an accepted blood substitute [8]) was used to generate these patterns. This blood was treated with aqueous acid citrate-dextrose (ACD) anticoagulant and kept refrigerated at all times prior to use (warmed up to 378C when required and blood used within 20 days from collection). Three replicate patterns were produced for each set. The first set of patterns (four in total) was created using an automated NE-300 Just InfusionTM syringe pump that pushed blood through plastic tubing, across the surface of two different ‘weapons’ (a cylindrical baton and a knife). Two or more blood
2.4. The classification scheme Element descriptors were combined to form a whole description, which collectively described a group of elements, which shared visible characteristics pertaining to their shape, margins and approximate dimensions. Each group was allocated a unique alphanumeric code, which was associated with observable element characteristics. This process ultimately led to the construction of a preliminary classification scheme. This classification scheme was further refined with the aid of a lay-participant workshop. The specific objectives for the workshop were (a) to test whether elements could be described in a consistent manner and without reference to a bloodstain pattern formation mechanism (b) to generate new descriptors for defining elements and (c) to determine whether element descriptions could then be used to reliably classify elements into meaningful groups. A total of eight individuals, primarily postgraduate students within the Forensic Science programme at The University of Auckland, New Zealand formed the lay-participant cohort. These participants had a science background, minimal BPA knowledge and no practical forensic experience. The group was chosen to provide a fresh look at terminology, by minimising the discipline jargon and potential biasing effect of existing methodologies. Ethics approval for the use of human participants was obtained from the University of Auckland Human Participants Ethics Committee, New Zealand (Ref: 8037). The workshop comprised five separate activities, which were conducted and completed simultaneously by all participants and with time restrictions imposed (1–2 min per task). A total of seven different element images were chosen for the various workshop tasks and were selected on the basis that they
R.M. Arthur et al. / Forensic Science International 257 (2015) 220–228
222
were an accurate representation of the element group description in question. The workshop activities were as follows: Activity 1: Participants were asked to draw sketches to reflect single element descriptors, taken from the preliminary classification scheme. Activity 2: Participants were asked to draw sketches in response to a more complete description of an element group. Activity 3: Participants were provided with a list of element descriptors selected from the preliminary classification scheme together with an image of an element. They were then asked to select those terms that they considered best described the appearance of the element. Activity 4: Participants were provided with images of three different element types. They were asked to describe the appearance of the element using common language descriptive terms of their own choosing. Activity 5: Images of different elements were provided together with the preliminary classification scheme. Participants were asked to use the scheme to select the most appropriate element group that best described the appearance of the element. Using responses from the workshop, the preliminary classification scheme was reviewed and modified and a visual aid termed ‘The Atlas of Elements’ was devised. This document consisted of modified element group descriptions and a combination of high speed and static images to characterise the element group. The role of the fluid dynamic principles of bloodstain formation was considered whilst studying the formation of the individual elements. In general, elements in a group were observed to follow a similar formation evolution. Understanding the fluid dynamics assisted with the identification of any observable and measurable element characteristics. This was accomplished by studying highspeed images of element formation.
The Atlas of Elements and a Glossary of Terms was provided to the BPA-trained participants, together with 17 images (at least two replicates per element group) representing the various elements. Participants were asked to classify the elements based on the groups designated by the Atlas. A successful classification was deemed to have occurred if the element was placed in the intended element group. 2.6. Characterising associations The digital images of the 15 laboratory-prepared bloodstain patterns (11 impact patterns, four-drip patterns) generated at the start of the study were used for the characterisation of associations. These digital images were converted into colour maps using Adobe Photoshop CS5, by inspecting each individual element in a single pattern and classifying it using the descriptions within the Atlas of Elements. A different colour was allocated for each element group and a coloured circle placed in the approximate location in the pattern for each element. Colour maps were created to visualise all element groups and to determine whether any associations could be identified for a particular pattern. Associations were characterised according to their spatial location and considered as either: Intra-element: The relationship within an element group, for example, between elements in Group 1A or elements in Group 1B or Inter-element: The relationship between element groups, for example, between Groups 1A and 3A. By creating colour maps, it was hypothesised that relationships within and between element groups could be visually assessed and their contribution to the distinctives of the overall pattern could be determined. 3. Results and discussion 3.1. Workshop
2.5. Validation of the atlas of elements A classification test was designed to test the use of the Atlas of Elements. This test was performed by an expert-participant group. The expert-participant group consisted of 17 participants. A total of 12 employed by ESR Ltd, were selected because they were BPA qualified and had the authority to carry out BPA reporting in casework. The remaining five participants, employed by the Netherlands Forensic Institute (NFI) held equivalent reporting qualifications.
The lay-participant workshop, composed of five different activities was used to validate and refine the descriptors that were generated to describe the physical appearance of elements. A complete analysis of the results of this workshop is reported in [10]. Fig. 1 shows the results of Activity 1, where participants drew sketches to reflect single element descriptors. The similarities in the drawn images suggest that a common understanding of well-defined element descriptors can be
Fig. 1. Participant drawings in response to the descriptor ‘smooth continuous boundary’.
R.M. Arthur et al. / Forensic Science International 257 (2015) 220–228
223
Fig. 2. Participant drawings in response to the description ‘oval base which tapers inwards to form a slim neck’.
achieved. It was evident that most participants tended to associate the terms ‘smooth continuous boundary’ with a circle shape. Fig. 2 shows the results of Activity 2, where participants drew sketches in response to the element description ‘an oval base which tapers inwards to form a slim neck.’ The sketches demonstrate a consistent understanding of the complete element description as well as the ability to recognise that elements can have different parts (i.e. a base and a neck have been drawn in different proportions). It is clear that element descriptions could be enhanced in the future by the addition of quantitative parameters. For example, the size of the neck and base could be expressed as a ratio so that this element could be better defined. Fig. 3 shows an example of the results of Activity 3 where a combination of terms from the preliminary classification system were selected by participants as best descriptors for the element in question. Simple shape-based terms, such as ‘circular’ and ‘uneven edges’ were used by all participants to describe the element shown in this example. This result provided the basis for incorporating these particular terms into the final element descriptions within the Atlas of Elements. Other terms, although less popular were selected as alternative descriptors (‘finger-like projections’ and ‘circular/oval droplet’) for this
element also. Activity 3 was most important for filtering out terms that could be misused. For example, the term ‘finger-like’ was used by 3 participants, however, upon review, this term was eliminated from the final classification system because it compared the appearance of the element to a physical object and therefore inherited the variability of the shape of that object. One of the main objectives of the workshop was to generate new descriptors for defining elements by giving participants an opportunity to describe the element using simple terms. This is illustrated in Fig. 4 which demonstrates the results of Activity 4, where participants described images using descriptive language of their own choosing. This activity gave participants more flexibility in the choice of terms and the opportunity to describe different parts of the element. Overall, a wide range of descriptive terms were generated for the elements provided. Participants generally had the tendency to describe elements in terms of regions. For example, as shown in Fig. 4, some terms clearly corresponded to the central part of the element (labelled Region 1) and others corresponded to the narrow part of the element (labelled Region 2). The results also showed that terms with similar meanings, such as ‘tail’ or ‘extension’ need to be well-defined if they were to be used to describe different features.
Fig. 3. Image of element and frequency of terms used by participants to describe it.
224
R.M. Arthur et al. / Forensic Science International 257 (2015) 220–228
Fig. 4. Examples of the results of Activity 4 showing two image of elements and the range of terms produced by participants. Terms are listed according to the main regions of the element.
The final objective of the workshop was to conduct a classification task in order to determine whether element descriptions could be used to reliably classify elements into meaningful groups. Examples of the results of Activity 5 are shown in Fig. 5.
For all three of the element examples shown in Fig. 5, 75% or more of the participants selected the group description intended to be associated with that element demonstrating that a high degree of classification precision is possible.
Fig. 5. Images of groups 3A, 3B and 4 elements provided to participants and their respective classification frequencies.
R.M. Arthur et al. / Forensic Science International 257 (2015) 220–228
225
Table 1 Excerpt from the Atlas of elements: Group 3B elements. Group 3B elements
Fluid dynamic mechanisms
High-speed video stills
Element attributes shape: Oval with a single projection referred to as a tail. Projection consists of a rounded tip and can appear partially transparent. Projection curves inwards on both sides between the tip and the base of the oval to form a slim neck.
1. Upon impact of a free-falling droplet onto the surface of a preexisting blood film, a sheet of blood (forming a distinct crown shape) forms due to a transfer of kinetic energy into surface energy (i.e. radial expansion).
2. Group 3B elements begin their early stages of formation from a secondary volume of blood that has detached from the original sheet. Sheet break-up and detachment occurs due to the growth of capillary waves travelling along the sheet.
Diagrammatic example
3. A droplet and its associated ligament detach from this secondary volume due to viscous shear and the surface tension force. 4. The surface tension force causes prolate and oblate oscillations which eventually die away leaving the blood in a spherical shaped droplet. Retraction of the ligament is seen as a result of the surface tension force minimising the surface area. 5. As the droplet impacts the surface at a low impact angle and with low energy, a transfer of kinetic to surface energy will drive spreading along the surface. A larger volume of blood is found at the base which causes it to appear raised. 6. Spreading is propagated along the surface. The surface tension force can be seen to act on the spherical tip.
Examples of variation
7. Retraction as a result of the surface tension force pulls some volume of blood to the centre of the element. The early formation of the projection is seen during this stage. This projection is limited in length because of the low energy of the impacting droplet. 8. Surface tension and viscous shear eventually slow down and halt the spreading motion. The single projection (tail) is more pronounced as spreading has eventually halted. The blood retains a raised profile. 9. Upon drying, the resulting element is an oval shape with a single projection referred to as a tail. Grid size: 10 mm 10 mm
3.2. Atlas of elements The classification scheme was further enhanced by a study of high-speed video images of examples of the formation of elements from each group. These images, along with static images of typical examples of resulting elements, were combined to form an ‘Atlas of Elements.’ An excerpt from this Atlas is shown in Table 1. The full Atlas in its present stage of development is presented in [10]. 3.3. Validation test of the atlas of elements The results of the validation test of the Atlas are shown in Fig. 6. Blue bars correspond to replicate one, that is the first example of an element in each group and the red bars correspond to replicate two. Fig. 6 indicates a high degree of precision was evident when the Atlas was used by BPA experts to classify elements. An overall mean classification success rate of 86% was achieved. All element examples in Groups 1B and 4 were successfully classified by all participants. Of interest, was the low success rate for one example of the elements in groups 1C, 2A and 3A. This suggests that some
elements are more variable in appearance than others and hence, more susceptible to misclassification. It is likely that these elements are also more difficult to describe unambiguously. To improve the classification rate, a larger database of example images may help analysts better understand the differences within these groups and a more in-depth study of their formation mechanism may shed light on these differences. Additionally, developing descriptions that better capture the essential distinctives of these groups might also help. Finally, changing the overall number of elements could have a bearing on the classification success rate 3.4. Characterising associations A total of 15 laboratory-generated bloodstain patterns were analysed for element associations. Associations were defined as relationships between elements in one group (intra-element) and between elements in different groups (inter-element). It was hypothesised that these associations could be used to distinguish between different bloodstain pattern types. The colour maps outlined in Figs. 7 and 8 document two examples of laboratorycreated patterns.
226
R.M. Arthur et al. / Forensic Science International 257 (2015) 220–228
Fig. 6. Classification results of elements by 17 BPA experts (*15 experts only) when using the Atlas of Elements to classify examples of the different elements, R1 and R2 refer to replicates 1 and 2, respectively.
Fig. 7 demonstrates a tightly clustered intra-element association between Group 1A (blue) elements in the lower central region of the pattern and a more dispersed grouping towards the periphery. These characteristics were reproducibly found in those patterns created with the in-house impact device. A scattered association between elements in Group 3B (black) also occurred in the impact patterns and was commonly found at the periphery of the pattern. Finally, all the impact patterns analysed in this study provided evidence of a scattered, mid-pattern association between Group 1B (pink) elements with a few isolated elements at the periphery. This mid-pattern association was also true for Group 3A (green) elements. There is a possibility that analysis of further patterns may reveal a consistent inter-element association between Groups 1B (pink) and 3A (green) elements for such patterns. Patterns generated as a result of dripping blood, tended to exhibit fewer elements dominated by Group 1A (blue) elements,
Fig. 7. Digital image of bloodstain pattern (left) and corresponding colour map (right). AC (area of convergence) has been specified. Groups; 1A (blue), 1B (pink), 1C (purple) 3A (green), 3B (black), 5 (red), 6 (yellow).
Fig. 8. Digital image of bloodstain pattern (left) and corresponding colour map (right). Group 1A (blue) Group 3A (green) around single group 2B (brown) element.
Fig. 9. Digital image of bloodstain pattern (left) on rear bumper of vehicle and corresponding colour map (right).
R.M. Arthur et al. / Forensic Science International 257 (2015) 220–228
which were typically randomly distributed within the pattern (see Fig. 8). To illustrate the possible application of this approach to casework, Fig. 9 exhibits an image of the rear of a vehicle with confirmed bloodstaining. A colour map of the elements identified in this pattern is also shown (Fig. 9). The analysis indicated that Group 1A (blue) elements, although fewer in number, were densely clustered near the centre of the pattern. A few Group 3B (black) elements were located at the periphery of the pattern and a mid-pattern scattering of Group 1B (pink) elements was noted. The apparent inter-element association between Groups 1B (pink) and 3A (green) elements was also noted for some closely spaced element pairs. Overall, the pattern exhibited some of the characteristic element associations noted for laboratory-created impact spatter patterns. 4. Conclusions Ultimately, BPA analysts aim to achieve two things. First, to reliably determine the likely mechanism(s) of formation of a bloodstain pattern and secondly, to be able to unambiguously articulate their conclusions and demonstrate how they were reached. To aid the analyst in these objectives, a novel, more objective classification methodology has been explored. The study provides a fresh look at the basics of bloodstain pattern analysis by focussing on the description of individual pattern components, termed ‘elements’ and analysing their associations within the pattern. In doing so, this method aimed to address some of the problems that have affected current classification methodologies. The results of this study show that, not only is it possible to break down a bloodstain pattern into components (elements), but it is possible to also describe these components using nonmechanism related nomenclature. Elements were grouped according to their appearance and how they were created, aided by the study of high-speed video images of element formation. Results indicated that it was possible to consistently describe elements and to distinguish between different element groups. There was also a tendency for certain groups of elements to be found in one specific bloodstain pattern. However, the validity of these results needs to be investigated further. In addition to this, it became apparent that a consideration of associations within and between element groups could form the basis for distinguishing different types of bloodstain patterns. The methodology presented in this study offers several potential advantages over current BPA methodologies. Firstly, the development of a common mechanism-free language for the sole purpose of describing the attributes of a pattern appears to be realistic. Secondly, a visual aid in the form of a prototype Atlas of Elements, that supports the analyst in his/her classification decision-making process, has been produced. The Atlas of Elements showcases a visual summary of the element-based classification system that was developed within this study. It is potentially a powerful tool that, with further development, could be used and easily understood by any expert to aid in the pattern classification process. The user can refer to a validated description that has incorporated simple mechanism-free terminology, static images comprised of examples of the element and finally a summary description of the fluid dynamics of element formation. The latter is supported by a series of in-depth high-speed video images, which documents significant detail of the formation stages of the element. It is evident that the method proposed within the Atlas of Elements not only offers the discipline a novel approach to classification but also capitalises on the use of imagery to support and strengthen expert interpretation. This should have the added benefit of making this an easy-to-articulate process in a court setting. Thus, bloodstain pattern classification could become a
227
demonstrably more objective process. Furthermore, the Atlas offers promise as a reference document that, if widely accepted, could bring added consistency to BPA. Finally the classification method proposed here is strongly rooted in the fundamental science of fluid dynamics. It should be noted that the concept proposed here remains a work in progress. Many more patterns need to be studied and the use of the classification method rigorously tested before its use in casework could be contemplated. The vital link between associated elements in a pattern and the mechanism of pattern formation needs to be established and validated. Furthermore, this study was limited to bloodstains on cardboard surfaces and only two of the major bloodstain pattern formation mechanisms (impact and drip). It is accepted that with changes in the target substrate and consideration of other pattern mechanisms, there will be other element types and associations to consider. This will further enrich the classification methodology proposed. Although the stated objectives of this study included the development of a more objective methodology, it should be clear that the classification methodology proposed still contains a subjective component. However, the proposed method lends itself to the addition of computer-vision based tools. Automated methods are already used in several forensic applications, such as, the comparison of striation marks resulting from fired bullets [11] and toolmark impressions [12] as well as for shoeprint retrieval and matching [13]. Statistical models such as principal component analysis have also been used in the study of accidental marks found on footwear [14,15]. It may also be possible to derive quantitative information about a bloodstain pattern using similar methods and use this information to develop a more automated and robust process for an element-based bloodstain pattern classification method. Acknowledgements The authors would like to acknowledge Dr. Douglas Elliot (ESR and School of Chemical Sciences, University of Auckland), Dr. John Buckleton (ESR) and Dr. Mark Jermy (University of Canterbury). Thank you to the participants from ESR and NFI for taking part in the experiments. This work was carried out in fulfilment of the Master of Science (Forensic Science) programme through the University of Auckland, New Zealand. The financial support of ESR through the ESR Masters Scholarship is sincerely appreciated. References [1] SWGSTAIN, Scientific working group on bloodstain pattern analysis: recommended terminology, Forensic Sci. Commun. 11 (2009). [2] T. Bevel, R.M. Gardner, Bloodstain Pattern Analysis with an Introduction to Crime Scene Reconstruction, CRC Press, Boca Raton: FL, USA, 2008. [3] T. Laber, P. Kish, M. Taylor, G. Owens, N. Osborne, J. Curran, Reliability Assessment of Current Methods in Bloodstain Pattern Analysis, National Institute of Justice, U.S. Department of Justice, 2014 Report No: 2010-DN-BX-K213. [4] J. Saviano, Articulating a concise scientific methodology for bloodstain pattern analysis, J. Forensic Identif. 55 (4) (2005) 461–470. [5] R. Gardner, Defining a methodology for bloodstain pattern analysis, J. Forensic Identif. 56 (4) (2006) 549–557. [6] Committee on Identifying the Needs of the Forensic Sciences Community, National Research Council, Strengthening Forensic Science in the United States: a Path Forward, National Academies Press, 2009 February. [7] L. Haber, R.N. Haber, Scientific validation of fingerprint evidence under Daubert, Law Prob. Ris. 7 (2008) 87–109. [8] M.A. Raymond, E.R. Smith, J. Liesegang, The physical properties of blood-forensic considerations, Sci. Justice 36 (1996) 153–160. [9] T. Stotesbury, The development of sensitivity tests for the creation and use of synthetic blood substitutes in bloodstain pattern analysis, in: Unpublished master’s thesis, The University of Auckland, New Zealand, 2012. [10] R. Arthur, The development of an element-based method for the reliable identification of bloodstain patterns, (Unpublished master’s thesis), The University of Auckland, New Zealand, 2013.
228
R.M. Arthur et al. / Forensic Science International 257 (2015) 220–228
[11] F.P. Leo´n, Automated comparison of firearm bullets, Forensic Sci. Int. 156 (2006) 40–50. [12] M. Baiker, I. Keereweer, R. Pieterman, E. Vermeij, J. van der Weerd, P. Zoon, Quantitative comparison of striated toolmarks, Forensic Sci. Int. 242 (2014) 186–199. [13] G. AlGarni, M. Hamiane, A novel technique for automatic shoeprint image retrieval, Forensic Sci. Int. 181 (2008) 10–14.
[14] N.D.K. Petraco, C. Gambino, T.A. Kubic, D. Olivio, N. Petraco, Statistical discrimination of footwear: a method for the comparison of accidentals on shoe outsoles inspired by facial recognition techniques, J. Forensic Sci. 55 (1) (2010) 34–41. [15] D. Sheets, S. Gross, G. Langenburg, P. Bush, M. Bush, Shape measurement tools in footwear analysis: a statistical investigation of accidental characteristics over time, Forensic Sci. Int. 232 (2013) 84–91.