Écrire un commentaire

Stainless steel slides has proved to be a game-changer for Raman spectroscopy, offering stronger signal intensity and high machine-learning accuracy at just four percent of the cost of traditional calcium fluoride slides. While calcium fluoride required long processing soaks to clear stubborn paraffin, stainless steel naturally avoided wax buildup and resisted breakage. By proving that cheap, rugged stainless steel can match or beat expensive specialty glass without sacrificing diagnostic power, this study removes a major practical bottleneck and paves the way for scalable, budget-friendly, artificial intelligence-driven digital pathology.

Major issues

• The authors appropriately note the limitation of the small cohort shown in the study. But it is worth emphasizing that validation on a larger, multi-center dataset will be necessary before broad clinical translation can be claimed.

• Since confusing epithelial misplacement with true cancer could lead to unnecessary treatments, how often do these machine learning models make false positive errors? Reporting the specific false positive rates (or specificity) would help show whether stainless steel is truly safe for clinical decision-making.

• If multiple slides came from the same patient, the model might accidentally learn from the same person twice, leading to data leakage. The authors should make sure their test splits are grouped strictly by patient.

• The authors notes that SS substrates may "induce paraffin melting under laser exposure" in addition to facilitating better chemical dewaxing. Could localized laser heating cause paraffin to melt and spread or coat the tissue micro-environment during acquisition?

• In this study, the tissue samples on the stainless steel and calcium fluoride slides were cut from different sections, minor changes in the tissue itself could affect the results. The authors should explain how they made sure these natural tissue differences didn't skew the comparison between the slides.

Minor issues

• Section 1 (Introduction): "there remains several elements" should be corrected to "there remain several elements".

• Section 3.3: "comparision" should be corrected to "comparison

• Ensure terms like SNV (Standard Normal Variate) and EM (Epithelial Misplacement) are fully spelled out upon their first appearance in the main text body.

• Ensure that all table and figure callouts in the text align cleanly with the final layout.

Competing interests

The author declares that they have no competing interests.

Use of Artificial Intelligence (AI)

The author declares that they did not use generative AI to come up with new ideas for their review.