Search

Tumors are typically analyzed as a single unit, despite their biologically heterogeneous nature. This limits correlations that can be drawn between regional variation and treatment outcome. Furthermore, despite the availability of high resolution 3-D medical imaging techniques, local outcomes, (e.g., tumor growth), are not easily measured. This paper proposes a method that uses streamlines to divide a 3-D region of interest (e.g., tumor) into units where local properties can be measured over the paths of growth. The parameters such as directional length and mean intensity can be measured locally at sequential time points and then compared. The method is evaluated on synthetic objects, simulated tumors, and medical images of brain tumors. The evaluations suggest that the method is suitable for mapping amorphous dynamic objects.

The feasibility of radiation therapy treatment planning using substitute computed tomography (sCT) generated from magnetic resonance images (MRIs) has been demonstrated by a number of research groups. One challenge with an MRI-alone workflow is the accurate identification of intraprostatic gold fiducial markers, which are frequently used for prostate localization prior to each dose delivery fraction. This paper investigates a template-matching approach for the detection of these seeds in MRI.Two different gradient echo T1 and T2* weighted MRI sequences were acquired from fifteen prostate cancer patients and evaluated for seed detection. For training, seed templates from manual contours were selected in a spectral clustering manifold learning framework. This aids in clustering "similar" gold fiducial markers together. The marker with the minimum distance to a cluster centroid was selected as the representative template of that cluster during training. During testing, Gaussian mixture modeling followed by a Markovian model was used in automatic detection of the probable candidates. The probable candidates were rigidly registered to the templates identified from spectral clustering, and a similarity metric is computed for ranking and detection.A fiducial detection accuracy of 95% was obtained compared to manual observations. Expert radiation therapist observers were able to correctly identify all three implanted seeds on 11 of the 15 scans (the proposed method correctly identified all seeds on 10 of the 15).An novel automatic framework for gold fiducial marker detection in MRI is proposed and evaluated with detection accuracies comparable to manual detection. When radiation therapists are unable to determine the seed location in MRI, they refer back to the planning CT (only available in the existing clinical framework); similarly, an automatic quality control is built into the automatic software to ensure that all gold seeds are either correctly detected or a warning is raised for further manual intervention.

Colonoscopy is performed by using a long endoscope inserted in the colon of patients to inspect the internal mucosa. During the intervention, clinicians observe the colon under bright light to diagnose pathology and guide intervention. We are developing a computer aided system to facilitate navigation and diagnosis. One essential step is to estimate the camera pose relative to the colon from video frames. However, within every colonoscopy video is a large number of frames that provide no structural information e.g. blurry or out of focus frames or those close to the colon wall. This hampers our camera pose estimation algorithm. To distinguish uninformative frames from informative ones, we investigated several features computed from each frame: corner and edge features matched with the previous frame, the percentage of edge pixels, and the mean and standard deviation of intensity in hue-saturation-value color space. A Random Forest classifier was used for classification. The method was validated on four colonoscopy videos that were manually classified. The resulting classification had a sensitivity of 75i¾ź% and specificity of 97i¾ź% for detecting uninformative frames. The proposed features not only compared favorably to existing techniques for detecting uninformative frames, but they also can be utilized for the camera navigation purpose.

Susceptibility-weighted imaging (SWI) is recognized as the preferred MRI technique for visualizing cerebral vasculature and related pathologies such as cerebral microbleeds (CMBs). Manual identification of CMBs is time-consuming, has limited reliability and reproducibility, and is prone to misinterpretation. In this paper, a novel computer-aided microbleed detection technique based on machine learning is presented: First, spherical-like objects (potential CMB candidates) with their corresponding bounding boxes were detected using a novel multi-scale Laplacian of Gaussian technique. A set of robust 3-dimensional Radon- and Hessian-based shape descriptors within each bounding box were then extracted to train a cascade of binary random forests (RF). The cascade consists of consecutive independent RF classifiers with low to high posterior probability constraints to handle imbalanced training sets (CMBs and non-CMBs), and to progressively improve detection rates. The proposed method was validated on 66 subjects whose CMBs were manually stratified into "possible" and "definite" by two medical experts. The proposed technique achieved a sensitivity of 87% and an average false detection rate of 27.1 CMBs per subject on the "possible and definite" set. A sensitivity of 93% and false detection rate of 10 CMBs per subject was also achieved on the "definite" set. The proposed automated approach outperforms state of the art methods, and promises to enhance manual expert screening. Benefits include improved reliability, minimization of intra-rater variability and a reduction in assessment time.

International audience; Recent developments of susceptibility weighted MR techniques have improved visualization of venous vasculature and underlying pathologies such as cerebral microbleed (CMB). CMBs are small round hypointense lesions on MRI images that are emerging as a potential biomarker for cerebrovascular disease. CMB manual rating has limited reliability, is time-consuming and is prone to errors as small CMBs can be easily missed or mistaken for venous crosssections. This paper presents a computer-aided detection technique that utilizes a novel cascade of random forest classifiers which are trained on robust Radon-based features with an unbalanced sample distribution. The training samples and their associated bounding box were acquired from a multiscale Laplacian of Gaussian technique with respect to their geometric characteristics. Validation results demonstrate that the current approach outperforms state of the art approaches with sensitivity of 92.04% and an average false detection rate of 16.84 per subject.

International audience; Automated detection of small, low level shapes such as circular/spherical objects in images is a challenging computer vision problem. For many applications, especially microbleed detection in Alzheimer's disease, an automatic pre-screening scheme is required to identify potential seeds with high sensitivity and reasonable specificity. A new method is proposed to detect spherical objects in 3D medical images within the multi-scale Laplacian of Gaussian framework. The major contributions are (1) breaking down 3D sphere detection into 1D line profile detection along each coordinate dimension, (2) identifying center of structures by normalizing the line response profile and (3) employing eigenvalues of the Hessian matrix at optimum scale for the center points to determine spherical objects. The method is validated both on simulated data and susceptibility weighted MRI images with ground truth provided by a medical expert. Validation results demonstrate that the current approach has higher performance in terms of sensitivity and specificity and is effective in detecting adjacent microbleeds, with invariance to intensity, orientation, translation and object scale.

PET-MRI fusion is widely used in oncology for early tumour diagnosis, localisation and monitoring of therapy effects. Automatic extraction of the lesions on PET images is desirable, but remains problematic. Manual segmentation of PET images is time consuming, and restricts the definition of the tumour extent to some arbitrary threshold. This can be sub-optimal in brain tumour for instance, where tumour is diffused by nature. Moreover, when the tracer uptake is not limited to the invaded regions, it becomes more difficult for an expert to define a precise contour. In this work, we propose a soft segmentation approach to automatically segment brain tumours in 18F-FDOPA PET images using a tumour growth model. This is based on extrapolating the tumour extent starting from tumour boundaries extracted from T1W MRI. A reaction-diffusion model is utilised for the extrapolation task to obtain tumour probability density. We evaluate our method on patient's PET/MRI images. The advantage of this method is that it is completely automatic and offers a soft segmentation of tumours in PET images.

3–D Eulerian–Eulerian CFD is used to simulate convective heat transfer in free falling particle curtains. Total heat loss for curtaining particles is compared to heat loss for isolated single particles. Spherical silica particles with density of 2,634 kg/m³ at 400 K (200 µm, 400 µm and 600 µm) flow at approximately 0.041 kg/s to 0.2 kg/s through a narrow slot in a rectangular box (0.45 m × 0.9 m × 0.225 m) filled with ambient air. The slot sizes through which the particles enter the rectangular box were 10 to 80 mm wide. Modifying the slot size at 0.041 kg/s for 400 µm particles can lead to 13% increases in rates of convective heat transfer per unit mass. A reduced order model was developed to predict the centreline temperatures of particles in the falling curtains. The drag coefficient in the ROM was varied to suit a range of particle sizes and mass flow rates.