Machine Learning Aids Diagnosis and Prognosis of Prostate Cancer Using MRI

Conventional magnetic resonance imaging (MRI) is a reliable tool for prognosis, diagnosis, active surveillance, and reducing the need for biopsy procedures in lower-risk prostate cancer patients. The integration of open-source data with machine learning models has created new opportunities to study disease progression and regression, particularly in the medical field. However, effectively incorporating machine learning in patient care poses several challenges, such as optimizing machine learning approaches for specific cancers, ensuring adequate specificity of training data for a particular medical condition, and more. In this context, generative adversarial networks (GANs) are being explored as a potential solution for generating high-quality synthetic data that accurately represent the clinical variability of a condition and can be applied to a range of imaging technologies, including PET, CT, MRI, ultrasound, and X-ray imaging in the brain, abdomen, and chest. However, while there has been some success, the use of GAN models is currently limited when it comes to accurately depicting the heterogeneity of complex diseases like prostate cancer.

A team of translational researchers at the University of Miami Health System (Miami, FL, USA) is leading the way in improving GAN tools for integration with diagnostic and prognostic tools in prostate cancer research. By requiring less data and patient follow-ups, GAN has the potential to revolutionize machine learning models and reduce healthcare costs and patient discomfort associated with repeat follow-ups. The goal is to use GAN's machine learning capabilities to generate digital images that learn from previous MRI images and clinical parameters, and predict disease progression or regression patterns.

The research team conducted a study using prostate MRIs and digital pathology from multiple sources as training data to develop a GAN model. With deep learning, they trained the model to segment the prostate boundary on both MRI and histology slices, which provide microscopic tissue structures. Experts with different levels of experience evaluated the generated images against traditional MRI images of the prostate. The study demonstrated that the prostate cancer MRIs produced using the model were of high quality. Deep learning segmentation helped remove images with significant distortion, indicating that this GAN machine learning model for prostate cancer imaging has promising implications for complex patient cases.

“Timely diagnosis and assessment of prognosis are challenges for prostate cancer, and this results in many deaths and increases [risk of disease progression],” said Himanshu Arora, Ph.D., assistant professor at Sylvester and the Desai Sethi Urology Institute at the Miller School of Medicine. “We cannot replace the human eye when it comes to medical decision-making. Still, the improvement in technologies could potentially assist radiation oncologists in making timely decisions.”

“Technically, the technology developed here is the first start to building more sophisticated models of ‘data augmentation’ where new digital images can be used in further analysis. This is an early phase of our study, but the outcomes are extremely promising,” Dr. Arora added.

Related Links:
University of Miami Health System

Visit expo >

Gold Member

Solid State Kv/Dose Multi-Sensor

AGMS-DM+

Compact C-Arm

Arcovis DRF-C S21

New

Full Field Digital Mammography Phantom

Mammo FFDM Phantom

Ultrasound Color LCD

U156W