by Christos Evangelou, MSc, PhD – Medical Writer and Editor
Analysis of whole-slide images (WSIs) by pathologists is an integral part of prostate cancer diagnosis. Therefore, the quality of WSIs plays a crucial role in the accuracy of prostate cancer diagnosis, and suboptimal staining of biopsy samples may lead to misdiagnosis or delayed diagnosis.
Stain normalization has been widely used to standardize color appearance and improve the chromatic features of WSIs. However, the effect of stain normalization on the accuracy of prostate cancer diagnosis remains unknown.
In a recent comparative study, researchers at the Polytechnic University of Turin and the University Hospital of Salerno analyzed the impact of stain normalization on perceived color quality, diagnostic confidence, and time required for diagnosis based on WSIs of prostate biopsies.1
The study showed that stain normalization increased color quality in digital images of prostate biopsies and reduced the time required for diagnosis based on prostate cancer biopsies, as assessed by two experts. In addition, stain normalization significantly increased diagnostic confidence in prostate cancer assessment.1
“The findings of our study suggest that stain normalization can play a valuable role in the routine practice of prostate cancer assessment, which can ultimately lead to better patient outcomes and quality of care,” said Massimo Salvi, PhD, assistant professor at the Polytechnic University of Turin and the first author of the study.
The report was published in the journal Cancers (Basel).
Study Rationale: Determining the Impact of Stain Normalization on Prostate Cancer Biopsies
Prostate cancer diagnosis and grading are based on the histopathological assessment of stained prostate biopsies. Nevertheless, the relatively high rates of inter- and intra-observer variability may influence the accuracy of diagnosis and tumor grading.
Perceived color variability can be reduced by using physical quality controls. Such controls include subjective assessments and comparisons of imaging data between laboratories. Nonetheless, color variations among tissue slides stained at different laboratories may hinder inter-laboratory comparisons and increase the perceived color variability.
Stain normalization has been widely used to help standardize the color appearance of WSIs using a reference image. However, the impact of stain normalization on the accuracy of pathological assessment of prostate cancer biopsies has not been determined.
Commenting on their motivation for this study, Dr. Salvi noted: “The study was conducted to address the challenges of inter- and intra-observer variability in diagnosing prostate cancer, which can be affected by suboptimal staining of biopsy samples. Stain normalization is a process that can standardize the color appearance of digital images, potentially improving the quality of images and facilitating the routine practice of prostate cancer assessment.”
Stain Normalization Increases Color Quality in Digital Images of Prostate Cancer Biopsies
To determine the impact of stain normalization on the accuracy of assessment of prostate biopsies, two pathology experts evaluated perceived color quality by assessing 93 original WSIs and their corresponding stain-normalized slides from patients with prostate cancer.1
For both experts, the median values of color quality were significantly higher for the stain-normalized slides than for their corresponding original WSIs (P < 0.0001). These findings suggest that stain normalization may improve perceived color quality in WSIs of prostate cancer biopsies.
Stain Normalization Reduces the Time Required for Diagnosis and Increases Diagnostic Confidence in Prostate Cancer
To determine whether stain normalization could improve prostate cancer diagnosis, two pathology experts tested the time required for diagnosis and diagnostic confidence by examining the original digital images and their corresponding stain-normalized WSIs. This comparison showed that stain normalization significantly reduced the time required for prostate cancer diagnosis.1
For the first expert, the average time for diagnosis was 69.9 ± 27.6 sec with stain-normalized WSIs and 77.9 ± 26.6 sec with the original WSIs (P < 0.0001). A similar difference in the time required for prostate cancer diagnosis was observed for the second expert (37.4 ± 24.4 sec vs. 52.7 ± 32.2 sec; P < 0.0001).1
A statistically significant increase in diagnostic confidence was also observed when stain normalization was used. For the first expert, confidence scores in prostate cancer diagnosis were 7.0 ± 1.8 for the original WSIs and 8.2 ± 1.2 for the corresponding stain-normalized WSIs (P < 0.0001). Similarly, confidence scores for the second expert were 6.2 ± 2.4 for original images and 7.2 ± 1.2 for the normalized ones (P = 0.00028).1
Novelty of Study and Future Work
When asked about the novelty of this study, Dr. Salvi said: “The most novel aspect of this work is the investigation of the impact of the stain normalization process on the assessment of prostate cancer biopsies from a pathologist’s perspective.”
The study’s focus on the evaluation of perceived color quality, diagnosis for the patient, diagnostic confidence, and time required for diagnosis by two pathology experts is unique.
“These parameters are crucial for the routine practice of prostate cancer assessment, and the study’s findings provide valuable insights into the potential benefits of stain normalization in improving the quality of images and facilitating the diagnosis of prostate cancer,” Dr. Salvi explained.
He also noted that their future research plans include further validation of the potential impact of stain normalization on the accuracy of prostate cancer diagnosis and patient outcomes through larger-scale studies and longitudinal follow-up. Additionally, the team plans to extend the same stain normalization process to other pathologies, including breast and lung cancer, as well as to other stains.
- Salvi M, Caputo A, Balmativola D, et al. Impact of Stain Normalization on Pathologist Assessment of Prostate Cancer: A Comparative Study. Cancers (Basel). 2023;15(5). doi:10.3390/cancers15051503