Search

PURPOSE: The goal of this work is to develop a method to identify women at high risk for having breast cancer that is easily missed in regular mammography screening. Such a method will provide a rationale for selecting women for adjunctive screening. It goes beyond current risk assessment models that are not specifically adapted to reduce the number of interval cancers. METHOD AND MATERIALS: From the Dutch breast cancer screening program we collected 109 cancers that were screen negative and subsequently appeared as interval cancers, and 327 age matched healthy controls. To obtain mammograms without signs of cancerous tissue, we took the contralateral mammograms. We developed a novel machine learning based method called convolutional sparse autoencoder (CSAE) to characterize mammographic texture. The CSAE was trained and tested on raw mammograms to separate interval cancers from controls in a five-fold cross validation. To assess the independency of the texture scores of breast density, density was determined for each image using Volpara. RESULTS: The odds ratios for interval cancer were 1.59 (95%CI: 0.76-3.32), 2.07 (1.02-4.20), and 3.14 (1.60-6.17) for quartile 2, 3 and 4 respectively, relative to quartile 1. Correlation between the texture scores and breast density was 0.59 (0.52-0.64). Breast density adjusted odds ratios, as determined with logistic regression, were 1.49 (0.71-3.13), 1.58 (0.75-3.33), and 1.97 (0.91-4.27). CONCLUSIONS: The CSAE texture score is independently associated with the risk of having a breast cancer that is missed in regular mammography screening. As such it offers opportunities to further enhance personalized breast cancer screening.

This study investigates whether measures of knee cartilage thickness can predict future loss of knee cartilage. A slow and a rapid progressor group was determined using longitudinal data, and anatomically aligned cartilage thickness maps were extracted from MRI at baseline. A novel machine learning framework was then trained using these maps. Compared to measures of mean cartilage plate thickness, group separation was increased by focusing on local cartilage differences. This result is central for clinical trials where inclusion of rapid progressors may help reduce the period needed to study effects of new disease-modifying drugs for osteoarthritis. (C) 2013 Elsevier Ltd. All rights reserved.

Osteoarthritis (OA) is a widespread, chronic joint disease for which there are currently no effective treatments beyond symptom relief. The lack of any approved disease modifying osteoarthritic drugs may partly be explained by insufficient disease understanding, but may also be tied to the absence of sensitive biomarkers for monitoring disease progression. This thesis investigates how subregional measures of cartilage thickness can be used to improve upon current imaging biomarkers. The first part of this investigation aims to discover discriminative areas in the cartilage using machine-learning techniques specifically developed to take advantage of the spatial nature of the problem. The methods were evaluated on data from a longitudinal study where detailed cartilage thickness maps were quantified from magnetic resonance images. The results showed that focal differences in cartilage thickness may be relevant for both OA diagnosis and for prediction of future cartilage loss. The second part of the thesis investigates spatial patterns of longitudinal cartilage thickness changes in healthy and OA knees. Based on our findings, we propose a new, conceptually simple biomarker that embraces the heterogeneous spatial cartilage changes that were observed in our study and in recent literature. The cartilage “Activity” marker is shown to have a state-of-the-art performance in separating healthy knees from OA knees and is also shown to predict knee replacement which is a clinically relevant endpoint for OA., Osteoarthritis (OA) is a widespread, chronic joint disease for which there are currently no effective treatments beyond symptom relief. The lack of any approved disease modifying osteoarthritic drugs may partly be explained by insufficient disease understanding, but may also be tied to the absence of sensitive biomarkers for monitoring disease progression. This thesis investigates how subregional measures of cartilage thickness can be used to improve upon current imaging biomarkers. The first part of this investigation aims to discover discriminative areas in the cartilage using machine-learning techniques specifically developed to take advantage of the spatial nature of the problem. The methods were evaluated on data from a longitudinal study where detailed cartilage thickness maps were quantified from magnetic resonance images. The results showed that focal differences in cartilage thickness may be relevant for both OA diagnosis and for prediction of future cartilage loss. The second part of the thesis investigates spatial patterns of longitudinal cartilage thickness changes in healthy and OA knees. Based on our findings, we propose a new, conceptually simple biomarker that embraces the heterogeneous spatial cartilage changes that were observed in our study and in recent literature. The cartilage “Activity” marker is shown to have a state-of-the-art performance in separating healthy knees from OA knees and is also shown to predict knee replacement which is a clinically relevant endpoint for OA.

Precise and robust whole brain, ventricle, and hippocampal atrophy measurements are important as they serve as biomarkers for Alzheimer’s disease. They are used as secondary outcomes in drug trials, and they correlate with the cognitive scores. When two successive scans are non-linearly aligned by registration, the volume change in a region of interest (ROI) is typically computed by Jacobian integration (JI), volumetric meshing (VM), or surface based methods like surface triangulation (ST) or surface flux (SF). JI and VM offer the possibility of a voxel-by-voxel atrophy measure for visualization or localization of atrophy and subsequent summing to an ROI measure of atrophy. ST and SF only offer whole ROI atrophy measures. JI and SF suffer from a lack of precision originating from respectively approximating a space and a time integral by a finite sum. VM suffers from a high computational burden and the ST from the lack of localization. In this paper we present the cube propagation (CP) algorithm having numerical precision as VM, offering the localization as JI and VM, but computational simplicity as the ST and SF. We demonstrate superior numerical precision to the the commonly used JI.