Stimulated raman histology allows for rapid pathologic examination of unprocessed, fresh prostate biopsies

Stimulated Raman Histology Interpretation by Artificial Intelligence Provides Near-Real-Time Pathologic Feedback for Unprocessed Prostate Biopsies

“…Moreover, variability in prostate histology interpretation, which includes SRH, complicates this process. 5 Thus, developing swift, accurate AI networks is essential for SRH implementation. An AI for H&E-stained prostate pathology, approved clinically, showed high sensitivity (97.7%) and specificity (99.3%).…”

“…[1][2][3] SRH rapidly creates high-resolution images of unprocessed, unlabeled, undyed fresh tissue, delivering both structural and functional histologic information of diagnostic quality for rapid prostate biopsy (PB) interpretation, thereby providing expedited histologic information within minutes. 4,5 Nonetheless, the presence of interobserver variability in genitourinary pathologists' interpretation of PB SRH remains a significant challenge, possibly acting as a barrier to widespread adoption. 5 The use of artificial intelligence (AI) to interpret SRH could overcome these limitations, facilitating procedural pathology in the diagnostic setting and accelerating tissue diagnosis.…”

“…4,5 Nonetheless, the presence of interobserver variability in genitourinary pathologists' interpretation of PB SRH remains a significant challenge, possibly acting as a barrier to widespread adoption. 5 The use of artificial intelligence (AI) to interpret SRH could overcome these limitations, facilitating procedural pathology in the diagnostic setting and accelerating tissue diagnosis. AI interpretation of central nervous system pathology using SRH has demonstrated noninferiority to pathologists in tumor diagnosis, significantly reducing the time needed for tissue diagnosis from over 30 minutes to just 2.5 minutes, and recently, enabling the identification of genetic mutations.…”

“…4,5 Nonetheless, the presence of interobserver variability in genitourinary pathologists' interpretation of PB SRH remains a significant challenge, possibly acting as a barrier to widespread adoption. 5…”

Editorial Comment

“…In this paper the authors move the needle further into the future, using Raman spectroscopy to create virtual histologic images of prostate biopsies, and training an artificial intelligence–based algorithm to recognize cancer with a high degree of concordance with expert review of standard hematoxylin and eosin slides. 1 When optimized, the system promises to deliver a dichotomous read of “cancer present or absent” in just a few minutes. As the authors highlight, such real-time feedback could be used to improve targeting and reduce the rate of false-negative biopsies (especially in those highly likely to have cancer based on MRI or other criteria), reduce the number of needle cores required to establish a diagnosis, and help identify margins for both focal therapy and radical prostatectomy.…”