A5087281080- AI in cancer detection
- Radiomics and Machine Learning in Medical Imaging
- Digital Radiography and Breast Imaging
- Medical Imaging Techniques and Applications
- Colorectal Cancer Screening and Detection
- Bladder and Urothelial Cancer Treatments
- Lung Cancer Diagnosis and Treatment
- Advanced X-ray and CT Imaging
- Cardiac Imaging and Diagnostics
- COVID-19 diagnosis using AI
- Medical Image Segmentation Techniques
- Urinary and Genital Oncology Studies
- Breast Lesions and Carcinomas
- Venous Thromboembolism Diagnosis and Management
- Global Cancer Incidence and Screening
- Coronary Interventions and Diagnostics
- Infrared Thermography in Medicine
- MRI in cancer diagnosis
- Image Retrieval and Classification Techniques
- Gene expression and cancer classification
- Urological Disorders and Treatments
- Artificial Intelligence in Healthcare and Education
- Prostate Cancer Diagnosis and Treatment
- Radiation Dose and Imaging
- Cerebrovascular and Carotid Artery Diseases
University of Michigan
2016-2025
Michigan United
2014-2025
Cleveland Clinic
2024
American Association of Physicists in Medicine
2019-2023
International Society for Optics and Photonics
2023
National Institutes of Health
2023
Michigan Medicine
2004-2022
Warde Medical Laboratory
2021
UPMC Health System
2019-2020
Swedish Medical Center
2019-2020
Develop a computer-aided detection (CAD) system for masses in digital breast tomosynthesis (DBT) volume using deep convolutional neural network (DCNN) with transfer learning from mammograms.A data set containing 2282 digitized film and mammograms 324 DBT volumes were collected IRB approval. The mass of interest on the images was marked by an experienced radiologist as reference standard. partitioned into training (2282 2461 230 views 228 masses) independent test (94 89 masses). For DCNN...
The authors are developing a computerized system for bladder segmentation in CT urography (CTU) as critical component computer-aided detection of cancer.A deep-learning convolutional neural network (DL-CNN) was trained to distinguish between the inside and outside using 160 000 regions interest (ROI) from CTU images. DL-CNN used estimate likelihood an ROI being ROIs centered at each voxel case, resulting map. Thresholding hole-filling were applied map generate initial contour bladder, which...
Transfer learning in deep convolutional neural networks (DCNNs) is an important step its application to medical imaging tasks. We propose a multi-task transfer DCNN with the aim of translating 'knowledge' learned from non-medical images diagnostic tasks through supervised training and increasing generalization capabilities DCNNs by simultaneously auxiliary studied this approach application: classification malignant benign breast masses. With Institutional Review Board (IRB) approval,...
In this paper, we developed a deep convolutional neural network (CNN) for the classification of malignant and benign masses in digital breast tomosynthesis (DBT) using multi-stage transfer learning approach that utilized data from similar auxiliary domains intermediate-stage fine-tuning. Breast imaging DBT, digitized screen-film mammography, mammography totaling 4039 unique regions interest (1797 2242 benign) were collected. Using cross validation, selected best six networks by varying level...
Abstract Cross-sectional X-ray imaging has become the standard for staging most solid organ malignancies. However, some malignancies such as urinary bladder cancer, ability to accurately assess local extent of disease and understand response systemic chemotherapy is limited with current approaches. In this study, we explored feasibility that radiomics-based predictive models using pre- post-treatment computed tomography (CT) images might be able distinguish between cancers without complete...
We are developing a computer‐aided diagnosis (CAD) system for lung nodule detection on thoracic helical computed tomography (CT) images. In the first stage of this CAD system, regions identified by k ‐means clustering technique. Each slice is classified as belonging to upper, middle, or lower part volume. Within each region, structures segmented again using weighted clustering. These may include true nodules and normal consisting mainly blood vessels. Rule‐based classifiers designed...
Digital tomosynthesis mammography (DTM) is a promising new modality for breast cancer detection. In DTM, projection‐view images are acquired at limited number of angles over angular range and the imaged volume reconstructed from two‐dimensional projections, thus providing three‐dimensional structural information tissue. this work, we investigated three representative reconstruction methods limited‐angle cone‐beam tomographic problem, including backprojection (BP) method, simultaneous...
We are developing a computer‐aided diagnosis (CAD) system to classify malignant and benign lung nodules found on CT scans. A fully automated was designed segment the nodule from its surrounding structured background in local volume of interest (VOI) extract image features for classification. Image segmentation performed with three‐dimensional (3D) active contour (AC) method. data set 96 (44 malignant, 52 benign) 58 patients used this study. The 3D AC model is based two‐dimensional addition...
Purpose To determine if digital breast tomosynthesis (DBT) performs comparably to mammographic spot views (MSVs) in characterizing masses as benign or malignant. Materials and Methods This IRB-approved, HIPAA-compliant reader study obtained informed consent from all subjects. Four blinded Mammography Quality Standards Act–certified academic radiologists individually evaluated DBT images MSVs of 67 (30 malignant, 37 benign) women (age range, 34–88 years). Images were viewed random order at...
The purpose of this work is to develop a computer-aided diagnosis (CAD) system differentiate malignant and benign lung nodules on CT scans. A fully automated was designed segment the nodule from its surrounding structured background in local volume interest (VOI) extract image features for classification. Image segmentation performed with 3D active contour method. initial obtained as boundary binary object generated by k-means clustering within VOI smoothed morphological opening. data set...
Grand challenges stimulate advances within the medical imaging research community; a competitive yet friendly environment, they allow for direct comparison of algorithms through well-defined, centralized infrastructure. The tasks two-part PROSTATEx Challenges (the Challenge and PROSTATEx-2 Challenge) are (1) computerized classification clinically significant prostate lesions (2) determination Gleason Grade Group in cancer, both based on multiparametric magnetic resonance images. incorporate...
Computer-aided detection and diagnosis (CAD) systems are increasingly being used as an aid by clinicians for interpretation of diseases. mark regions image that may reveal specific abnormalities to alert these during interpretation. provide assessment a disease using image-based information alone or in combination with other relevant diagnostic data decision support developing their diagnoses. While CAD commercially available, standardized approaches evaluating reporting performance have not...
Radiomics is to provide quantitative descriptors of normal and abnormal tissues during classification prediction tasks in radiology oncology. Quantitative Imaging Network members are developing radiomic "feature" sets characterize tumors, general, the size, shape, texture, intensity, margin, other aspects imaging features nodules lesions. Efforts ongoing for an ontology describe lung nodules, with main classes consisting local global shape descriptors, texture-based features, which based on...
The purpose of this work is to describe the LUNGx Challenge for computerized classification lung nodules on diagnostic computed tomography (CT) scans as benign or malignant and report performance participants’ methods along with that six radiologists who participated in an observer study performing same task dataset. provided sets calibration testing scans, established a assessment process, created infrastructure case dissemination result submission. Ten groups applied their own 73 (37 36...
To evaluate the feasibility of using an objective computer-aided system to assess bladder cancer stage in CT Urography (CTU).
Deep learning models are highly parameterized, resulting in difficulty inference and transfer for image recognition tasks. In this work, we propose a layered pathway evolution method to compress deep convolutional neural network (DCNN) classification of masses digital breast tomosynthesis (DBT). The objective is prune the number tunable parameters while preserving accuracy. first stage learning, 19 632 augmented regions-of-interest (ROIs) from 2454 mass lesions on mammograms were used train...
Rapid advances in artificial intelligence (AI) and machine learning, specifically deep learning (DL) techniques, have enabled broad application of these methods health care. The promise the DL approach has spurred further interest computer-aided diagnosis (CAD) development applications using both "traditional" newer DL-based methods. We use term CAD-AI to refer this expanded clinical decision support environment that uses traditional AI Numerous studies been published date on tools for...
Abstract Context Inhibition of the neonatal fragment crystallizable receptor (FcRn) reduces pathogenic thyrotropin antibodies (TSH-R-Ab) that drive pathology in thyroid eye disease (TED). Objective We report first clinical studies an FcRn inhibitor, batoclimab, TED. Design Proof-of-concept (POC) and randomized, double-blind placebo-controlled trials. Setting Multicenter. Participants Patients with moderate-to-severe, active Intervention In POC trial, patients received weekly subcutaneous...
Abstract The adoption of artificial intelligence (AI) tools in medicine poses challenges to existing clinical workflows. This commentary discusses the necessity context-specific quality assurance (QA), emphasizing need for robust QA measures with control (QC) procedures that encompass (1) acceptance testing (AT) before use, (2) continuous QC monitoring, and (3) adequate user training. discussion also covers essential components AT QA, illustrated real-world examples. We highlight what we see...