Ceyhun Burak Akgül

Content-based image retrieval in radiology: current status and future directions.

Diagnostic radiology requires accurate interpretation of complex signals in medical images. Content-based image retrieval (CBIR) techniques could be valuable to radiologists in assessing medical images by identifying similar images in large archives that could assist with decision support. Many advances have occurred in CBIR, and a variety of systems have appeared in nonmedical domains; however, permeation of these methods into radiology has been limited. Our goal in this review is to survey CBIR methods and systems from the perspective of application to radiology and to identify approaches developed in nonmedical applications that could be translated to radiology. Radiology images pose specific challenges compared with images in the consumer domain; they contain varied, rich, and often subtle features that need to be recognized in assessing image similarity. Radiology images also provide rich opportunities for CBIR: rich metadata about image semantics are provided by radiologists, and this information is not yet being used to its fullest advantage in CBIR systems. By integrating pixel-based and metadata-based image feature analysis, substantial advances of CBIR in medicine could ensue, with CBIR systems becoming an important tool in radiology practice.

3D Model Retrieval Using Probability Density-Based Shape Descriptors

We address content-based retrieval of complete 3D object models by a probabilistic generative description of local shape properties. The proposed shape description framework characterizes a 3D object with sampled multivariate probability density functions of its local surface features. This density-based descriptor can be efficiently computed via kernel density estimation (KDE) coupled with fast Gauss transform. The non-parametric KDE technique allows reliable characterization of a diverse set of shapes and yields descriptors which remain relatively insensitive to small shape perturbations and mesh resolution. Density-based characterization also induces a permutation property which can be used to guarantee invariance at the shape matching stage. As proven by extensive retrieval experiments on several 3D databases, our framework provides state-of-the-art discrimination over a broad and heterogeneous set of shape categories.

A Decision Forest Based Feature Selection Framework for Action Recognition from RGB-Depth Cameras

In this paper, we present an action recognition framework leveraging data mining capabilities of random decision forests trained on kinematic features. We describe human motion via a rich collection of kinematic feature time-series computed from the skeletal representation of the body in motion. We discriminatively optimize a random decision forest model over this collection to identify the most effective subset of features, localized both in time and space. Later, we train a support vector machine classifier on the selected features. This approach improves upon the baseline performance obtained using the whole feature set with a significantly less number of features (one tenth of the original). On MSRC-12 dataset (12 classes), our method achieves 94% accuracy. On the WorkoutSU-10 dataset, collected by our group (10 physical exercise classes), the accuracy is 98%. The approach can also be used to provide insights on the spatiotemporal dynamics of human actions.

Extraction of cognitive activity related waveforms from functional near infrared signals

We address the problem of prototypical waveform extraction in cognitive experiments using functional near-infrared spectroscopy (fNIRS) signals. These waveform responses are evoked with visual stimuli provided in an oddball type experimental protocol. As the statistical signal-processing tool, we consider the linear signal space representation paradigm and use independent component analysis (ICA). The assumptions underlying ICA is discussed in the light of the signal measurement and generation mechanisms in the brain. The ICA-based waveform extraction is validated based both on its conformance to the parametric brain hemodynamic response (BHR) model and to the coherent averaging technique. We assess the intra-subject and inter-subject waveform and parameter variability.

Density-based 3D shape descriptors

We propose a novel probabilistic framework for the extraction of density-based 3D shape descriptors using kernel density estimation. Our descriptors are derived from the probability density functions (pdf) of local surface features characterizing the 3D object geometry. Assuming that the shape of the 3D object is represented as a mesh consisting of triangles with arbitrary size and shape, we provide efficient means to approximate the moments of geometric features on a triangle basis. Our framework produces a number of 3D shape descriptors that prove to be quite discriminative in retrieval applications. We test our descriptors and compare them with several other histogram-based methods on two 3D model databases, Princeton Shape Benchmark and Sculpteur, which are fundamentally different in semantic content and mesh quality. Experimental results show that our methodology not only improves the performance of existing descriptors, but also provides a rigorous framework to advance and to test new ones.

Similarity Learning for 3D Object Retrieval Using Relevance Feedback and Risk Minimization

We introduce a similarity learning scheme to improve the 3D object retrieval performance in a relevance feedback setting. The proposed algorithm relies on a score fusion approach that linearly combines elementary similarity scores originating from different shape descriptors into a final similarity function. Each elementary score is modeled in terms of the posterior probability of a database item being relevant to the user-provided query. The posterior parameters are learned via off-line discriminative training, while the optimal combination of weights to generate the final similarity function is obtained by on-line empirical ranking risk minimization. This joint use of on-line and off-line learning methods in relevance feedback not only improves the retrieval performance significantly as compared to the totally unsupervised case, but also outperforms the standard support vector machines based approach. Experiments on several 3D databases, including the Princeton Shape Benchmark, show also that the proposed algorithm has a better small sample behavior.

Similarity score fusion by ranking risk minimization for 3D object retrieval

In this work, we introduce a score fusion scheme to improve the 3D object retrieval performance. The state of the art in 3D object retrieval shows that no single descriptor is capable of providing fine grain discrimination required by prospective 3D search engines. The proposed fusion algorithm linearly combines similarity information originating from multiple shape descriptors and learns their optimal combination of weights by minimizing the empirical ranking risk criterion. The algorithm is based on the statistical ranking framework [CLV07], for which consistency and fast rate of convergence of empirical ranking risk minimizers have been established. We report the results of ontology-driven and relevance feedback searches on a large 3D object database, the Princeton Shape Benchmark. Experiments show that, under query formulations with user intervention, the proposed score fusion scheme boosts the performance of the 3D retrieval machine significantly.

Spectral Analysis of Event-Related Hemodynamic Responses in Functional Near Infrared Spectroscopy

The goal of this paper is to design experiments that confirm the evidence of cognitive responses in functional near infrared spectroscopy and to establish relevant spectral subbands. Hemodynamic responses of brain during single-event trials in an odd-ball experiment are measured by functional near infrared spectroscopy method. The frequency axis is partitioned into subbands by clustering the time-frequency power spectrum profiles of the brain responses. The predominant subbands are observed to confine the 0–30 mHz, 30–60 mHz, and 60–330 mHz ranges. We identify the group of subbands that shows strong evidence of protocol-induced periodicity as well as the bands where good correlation with an assumed hemodynamic response models is found.

Multivariate Density-Based 3D Shape Descriptors

We address the 3D object retrieval problem using multivariate density-based shape descriptors. Considering the fusion of first and second order local surface information, we construct multivariate features up to five dimensions and process them by the kernel density estimation methodology to obtain descriptor vectors. We can compute these descriptors very efficiently using the fast Gauss transform algorithm. We also make use of descriptor level information fusion by concatenating descriptor vectors to increase their discrimination power further. To render the resulting descriptors storage-wise efficient, we develop two analytical tools, marginalization and probability density suppression, for descriptor dimensionality reduction. The experiments on two different databases, Princeton Shape Benchmark and Sculpteur, show that, boosted with both feature level and descriptor level information fusion, and powered with fast computational schemes, the density-based shape description framework enables effective and efficient 3D object retrieval.

Automated diagnosis of Alzheimer's disease using image similarity and user feedback

In this work, we present a learning framework to help early diagnosis of Alzheimers disease (AD) from magnetic resonance images using visual similarity and user feedback. Our approach relies on a nearest neighbor (NN) procedure where the similarity measure is obtained via on-line supervised learning. This framework differs from standard classification based medical diagnosis in that learning is always carried out on-line with a small training set, much like in relevance feedback-driven retrieval. We propose two alternative approaches to learn the similarities between cases. While the first approach indirectly employs the distance to support vector machine decision boundary as a similarity measure, the second one aims at directly finding a similarity function based on the minimization of the empirical ranking risk. Several experiments on Open Access Series of Imaging Studies neuroimaging database establish that, even with weak global visual descriptors and small training sets, this framework has better diagnostic performance than standard classification based approaches and it also enjoys a certain degree of robustness against incorrect relevance judgments.