Bogdan J. Matuszewski

Assessment of algorithms for mitosis detection in breast cancer histopathology images.

The proliferative activity of breast tumors, which is routinely estimated by counting of mitotic figures in hematoxylin and eosin stained histology sections, is considered to be one of the most important prognostic markers. However, mitosis counting is laborious, subjective and may suffer from low inter-observer agreement. With the wider acceptance of whole slide images in pathology labs, automatic image analysis has been proposed as a potential solution for these issues. In this paper, the results from the Assessment of Mitosis Detection Algorithms 2013 (AMIDA13) challenge are described. The challenge was based on a data set consisting of 12 training and 11 testing subjects, with more than one thousand annotated mitotic figures by multiple observers. Short descriptions and results from the evaluation of eleven methods are presented. The top performing method has an error rate that is comparable to the inter-observer agreement among pathologists.

Standardized evaluation framework for evaluating coronary artery stenosis detection, stenosis quantification and lumen segmentation algorithms in computed tomography angiography

Though conventional coronary angiography (CCA) has been the standard of reference for diagnosing coronary artery disease in the past decades, computed tomography angiography (CTA) has rapidly emerged, and is nowadays widely used in clinical practice. Here, we introduce a standardized evaluation framework to reliably evaluate and compare the performance of the algorithms devised to detect and quantify the coronary artery stenoses, and to segment the coronary artery lumen in CTA data. The objective of this evaluation framework is to demonstrate the feasibility of dedicated algorithms to: (1) (semi-)automatically detect and quantify stenosis on CTA, in comparison with quantitative coronary angiography (QCA) and CTA consensus reading, and (2) (semi-)automatically segment the coronary lumen on CTA, in comparison with experts manual annotation. A database consisting of 48 multicenter multivendor cardiac CTA datasets with corresponding reference standards are described and made available. The algorithms from 11 research groups were quantitatively evaluated and compared. The results show that (1) some of the current stenosis detection/quantification algorithms may be used for triage or as a second-reader in clinical practice, and that (2) automatic lumen segmentation is possible with a precision similar to that obtained by experts. The framework is open for new submissions through the website, at http://coronary.bigr.nl/stenoses/.

Hi4D-ADSIP 3-D dynamic facial articulation database

The face is an important medium used by humans to communicate, and facial articulation also reflects a persons emotional and awareness states, cognitive activity, personality or wellbeing. With the advances in 3-D imaging technology and ever increasing computing power, automatic analysis of facial articulation using 3-D sequences is becoming viable. This paper describes Hi4D-ADSIP - a comprehensive 3-D dynamic facial articulation database, containing scans with high spatial and temporal resolution. The database is designed not only to facilitate studies on facial expression analysis, but also to aid research into clinical diagnosis of facial dysfunctions. The database currently contains 3360 facial sequences captured from 80 healthy volunteers (control subjects) of various age, gender and ethnicity. The database has been validated using psychophysical experiments used to formally evaluate the accuracy of the recorded expressions. The results of baseline automatic facial expression recognition methods using Eigen- and Fisher-faces are also presented alongside some initial results obtained for clinical cases. This database is believed to be one of the most comprehensive repositories of facial 3-D dynamic articulations to date. The extension of this database is currently under construction aiming at building a comprehensive repository of representative facial dysfunctions exhibited by patients with stroke, Bells palsy and Parkinsons disease.

A Novel Medical Image Segmentation Method using Dynamic Programming

A novel method is proposed to segment objects in medical images whose boundaries can be described as closed curves. Based on an image with the enhanced boundary of an object of interest, the segmentation method consists of three key steps, namely, the polar transformation, dynamic programming and curve fitting. A 3D object in volumetric data can be segmented on a slice-by-slice basis by only specifying one point inside the 3D object of interest as the pole for the polar transformation. The method is also shown to be able to segment objects with very weak boundaries.

An Efficient Feature Based Matching Algorithm for Stereo Images

A novel efficient feature based stereo matching algorithm is presented in this paper. The proposed method links the detected feature points into chains and the matching process is achieved by comparing some of the feature points from different chains. A matching score based on 2 dimensional normalised cross correlation (2D NCC) is used to determine whether feature points are well matched to construct a feature correspondence. This process improves the reliability and the efficiency of the algorithm by concentrating on matching corresponding chains. The proposed method is tested and validated using real scenes and synthetic data images. Experimental results indicate that this novel algorithm is more reliable especially for images in which a number of vertical features are detected. It also compares well with existing methods in terms of speed of execution

Reduction of motion artefacts in on-board cone beam CT by warping of projection images.

OBJECTIVE We describe the development and testing of a motion correction method for flat panel imager-based cone beam CT (CBCT) based on warping of projection images. METHODS Markers within or on the surface of the patient were tracked and their mean three-dimensional (3D) position calculated. The two-dimensional (2D) cone beam projection images were then warped before reconstruction to place each marker at the projection from its mean 3D position. The motion correction method was tested using simulated cone beam projection images of a deforming virtual phantom, real CBCT images of a moving breast phantom and clinical CBCT images of a patient with breast cancer and another with pancreatic cancer undergoing radiotherapy. RESULTS In phantom studies, the method was shown to greatly reduce motion artefacts in the locality of the radiotherapy target and allowed the true surface shape to be accurately recovered. The breast phantom motion-compensated surface was within 1 mm of the true surface shape for 90% of surface points and greater than 2 mm from the true surface at only 2% of points. Clinical CBCT images showed improved image quality in the locality of the radiotherapy target after motion correction. CONCLUSION The proposed method is effective in reducing motion artefacts in CBCT images.

Comparative Validation of Polyp Detection Methods in Video Colonoscopy: Results From the MICCAI 2015 Endoscopic Vision Challenge

Colonoscopy is the gold standard for colon cancer screening though some polyps are still missed, thus preventing early disease detection and treatment. Several computational systems have been proposed to assist polyp detection during colonoscopy but so far without consistent evaluation. The lack of publicly available annotated databases has made it difficult to compare methods and to assess if they achieve performance levels acceptable for clinical use. The Automatic Polyp Detection sub-challenge, conducted as part of the Endoscopic Vision Challenge (http://endovis.grand-challenge.org) at the international conference on Medical Image Computing and Computer Assisted Intervention (MICCAI) in 2015, was an effort to address this need. In this paper, we report the results of this comparative evaluation of polyp detection methods, as well as describe additional experiments to further explore differences between methods. We define performance metrics and provide evaluation databases that allow comparison of multiple methodologies. Results show that convolutional neural networks are the state of the art. Nevertheless, it is also demonstrated that combining different methodologies can lead to an improved overall performance.

Choice of similarity measure, likelihood function and parameters for histogram based particle filter tracking in CCTV grey scale video

The choice of particle filter dissimilarity distance measures and likelihood functions is considered in the context of object tracking in grey scale CCTV video. The geometrical interpretation of the Bhattacharyya coefficient and distance is reviewed and the relationships between the Bhattacharyya, Matusita, histogram intersection and @g^2 distances are examined. It is argued that as long as the likelihood function satisfies certain criteria its analytical form is not critical in the stated tracking context. This is demonstrated through an experimental comparison between the use of the standard Bhattacharyya distance/Gaussian likelihood combination and the potentially computationally simpler histogram intersection distance/triangular likelihood combination in particle filter tracking sequences. It is shown that the differences between the approaches are marginal when the likelihood criteria are applied. Whilst the analysis was focused on a specific application and context, we suggest that the findings will be of value to particle filter tracking in general.

Boundary delineation in prostate imaging using active contour segmentation method with interactively defined object regions

Active contour methods are often methods of choice for demanding segmentation problems, yet segmentation of medical images with complex intensity patterns still remains a challenge for these methods. This paper proposes a method to incorporate interactively specified foreground/background regions into the active model framework while keeping the user interaction to the minimum. To achieve that, the proposed functional to be minimized includes a term to encourage active contour to separate the points close to the specified foreground region from the points close to the specified background region in terms of geodesic distance. The experiments on multi-modal prostate images demonstrate that the proposed method not only can achieve robust and accurate results, but also provides an efficient way to interactively improve the results.

Deformable image registration

The paper describes a novel deformable data registration algorithm. The proposed method can be seen as a tradeoff between the landmark and intensity driven data registration techniques. The algorithm is fast, robust and offers high registration accuracy. The algorithm enables to include complex constraints on the allowable data deformations. It is shown that some of these constraints permit to treat part of the data as rigid objects or restrict its deformation to a predefined shape. The main focus of the paper is put on the description of the elastic deformation model used in the method. The performance of this new technique is illustrated using simulated as well as real computed tomography (CT) images.