1. An Automated Diagnostic Workflow for the Diagnosis of Pulmonary Embolism in the Presence of CTPA and Clinical Evidence

Our goal in this project is to develop an integrated computer aided decision support (CAD) system which, when introduced into the Physician's workflow as a "second reader" improves the diagnosis of acute PE with the following benefits: more accurate diagnosis, resulting in fewer complications from unnecessary anticoagulation therapy, and reduced costs and radiation exposure, from the elimination of unnecessary follow-up imaging studies.

Initial Funding: IUPUI RSFG

2. A Machine Learning Approach to Label-free Detection of Bacterial Pathogens using Laser Light Scattering

Technologies for rapid detection and classification of bacterial pathogens are crucial for securing the food supply. A light-scattering sensor recently developed for real-time detection and identification of colonies of multiple pathogens has shown great promise for distinguishing bacteria cultures at the genus and species level for Listeria, Staphylococcus, Salmonella, Vibrio, and Escherichia. Unlike traditional testing methods, this new technology does not require a labeling reagent or biochemical processing. The classification approach currently used with this technology relies on supervised learning. For an accurate detection and classification of bacterial pathogens, the training library used to train the classifier should consist of samples of all possible forms of the pathogens. Construction of such a training library is impractical if not impossible due to the high mutation rate that characterizes some of the infectious agents. In this project we propose to advance this sensor technology to allow for the detection of new classes/subclasses of bacteria, which do not exist in the training library. Learning with a nonexhaustive training library is an ill-defined problem. We design a two stage classification scheme to alleviate this problem. The first stage, i.e. detection, identifies whether the bacteria sample belongs to one of the subclasses in the training library or a yet unseen and thus unrepresented subclass. If the former is true, the sample is fed to the second stage, i.e. classification, where it is classified to one of the existing subclasses. If the latter is true, an alert is raised and the sample is saved for follow-up analysis. Benefit for Public Health: Successful implementation of this project will allow for a label-free detection and identification of food pathogens and their mutated subclasses not yet seen earlier. This will reduce the number of food related outbreaks and will help secure our food supply.

Funding pending

3. Predictive Mathematical Models of Intraductal Breast Lesions

In the diagnosis of early-stage breast cancer, some of the intraductal lesions usually pose diagnostic difficulties, the most frequently encountered of which is whether a lesion represents epithelial hyperplasia or carcinoma in situ. The main difficulty in this differential diagnosis is due to an intermediate category of intraductal proliferations between usual epithelial hyperplasia(UDH) and Ductal Carcinoma in situ (DCIS), called Atypical Ductal Hyperplasia(ADH). ADH is most commonly considered as a proliferative lesion that meets some but not all criteria for low-grade, non-comedo type DCIS (LG-DCIS). Even though neither UDH nor ADH are precursors to carcinoma in situ or invasive carcinoma, women with ADH, is prone to four to five times increased risk of invasive breast carcinoma compared to those in the general population.
The criteria used to define ADH in histopathology is far from being perfect and the interobserver variability in the classification of all the “borderline” intraductal lesions, i.e. UDH vs ADH and ADH vs LG-DCIS, remains extremely significant. In a study that presents the results of a systematic review of published studies involving 39,560 adult patients who were evaluated for breast symptoms, it is found that the initial diagnosis of ADH made by core biopsy, changed in 42% of the patients as a result of excisional biopsy. Most changes were to DCIS or invasive cancer, although approximately 20% changed to more benign diagnoses. The diagnosis of borderline intraductal lesions requires extracting and analyzing a large number of quantitative features at multiple resolutions, which we believe is a task that can be more reliably and accurately performed with assistance from predictive mathematical systems. In this project, we propose to develop a predictive system which, when introduced into the pathologists’ workflow improves the diagnosis of borderline intraductal lesions, thereby provide the following health benefits: more accurate diagnosis of early stage breast cancer, resulting in fewer complications, less discomfort for the patient and reduced cost from the elimination of unnecessary resections. We will develop two predictive systems, one for each of H&E and caPCNAab-stained breast tissue samples and then conduct a reader study involving at least three pathologists to clinically validate each system in the diagnostic workflow as a computer-based image analysis tool for providing “second opinion” to the pathologists.

Funding pending