Richard Torres, MD, MS

For several years, my research laboratory has worked at the interface of engineering and medicine, applying new advances to important clinical laboratory problems. Specifically, a principal focus of my professional research interest is in translating technological developments in optics and image analysis into practical approaches to problems in diagnostic pathology. The impetus for this work has been the first-hand experience of enormous technical limitations affecting routine practice, for which technology offers exciting opportunities for improvement. The laboratory continually engages clinical practitioners and cutting-edge investigator colleagues to identify which evolving methods, techniques, and approaches can be readily molded into practical tools and have the best opportunities for significant impact on the understanding, diagnosis, and management of disease. We apply new techniques in a range of engineering fields such as rapid prototyping, electronics design, modern machine learning applications, and optics research, complemented by clinical experience which informs the understanding of pathology methods, diagnostic challenges, and clinical relevance. A principal area of research is a new method of tissue preparation and advanced microscopy that seeks to transform the manner in which pathologists render diagnoses for a whole host of diseases, including cancer of all types and renal disease.

The work that we have done over the last decade for the development of a practical technique and instrument for non-destructive three-dimensional histology is of immense potential impact to both human diagnostics and animal studies of pathophysiology of disease. The success has been made possible by focused innovations in a variety of areas. Together with long-time collaborator Michael Levene and in concert with several other groups, we pioneered use of benzyl alcohol/benzyl benzoate (BABB) clearing for multiphoton microscopy. We developed strategies incorporating chemical modificaitons to speed up clearing using BABB, reducing processing time of small samples from days to a few hours. Early on we recognized the utility of a combined nuclear/protein fluorescent dye approach for visual reproduction of common pathology stains and identified the DAPI/eosin combination subsequently adopted by others. We designed and built the fastest high-resolution multiphoton microscope in the world, achieving light-microscopy level quality quickly and at depth, enabling heretofore largely impossible large-scale high-resolution reconstructions of animal tissue for investigative projects as well as practical clinical application for diagnostic pathology. And we have developed specialized efficient software for visualization of these large-scale three dimensional image stacks.

These techniques offer numerous potential advantages over traditional methods in terms of completeness of analysis, preservation of tissue for advanced molecular studies, speed and cost savings, and examination of important parameters such as architecture and abnormal growth, all while maintaining the compatibility with traditional histologic analysis.

We are continuing to both demonstrate the clinical and investigational prowess of the new technologies as well as further develop performance capabilities. Clinical trials of the application of the methods and techniques are well-advanced in prostate cancer and renal disease, and we have begun examining the increased diagnostic capabilities of multiphoton microscopy of bone marrow aspirate specimens for diagnosis and pathophysiology research studies.

Additionally, like other disciplines, we have been exploiting large data initiatives in pathology with are poised to benefit immensely from high quality digital data. Both three-dimensional imaging with digitization of morphologic data and genomic sequencing result in generation of vast amounts of information which require computing management and analysis tools in order to yield utility. Work done with current and former trainees with specialized knowledge of powerful new computing tools for imaging and genomic data management are the source of on-going computing projects. These developments are demonstrating that newest image pattern recognition techniques based on convolutional neural networks and python programming modules can be employed to aid in the accurate characterization of pathology image data. We are extending our data management and analysis techniques to a variety of anatomic and clinical pathology projects.