Validation of Artificial Intelligence–enhanced Stimulated Raman Histopathology for Intraoperative Margin Assessment During Robot-assisted Radical Prostatectomy: Preliminary Results from the ROBOSPEC Study
Bibliographic record
Abstract
Our preliminary findings suggest that simulated Raman histology (SRH) with artificial intelligence assistance can support intraoperative decision-making and demonstrate potential for SRH feasibility in surgical margin assessment. While these results are encouraging, ongoing direct comparisons to frozen sections stained with hematoxylin and eosin and to paraffin-embedded sections are essential to further validate the findings and enhance the reliability of this strategy. Stimulated Raman histology (SRH) offers promising near–real-time tissue visualization for intraoperative pathology assessment. We present preliminary results from the ROBOSPEC study, with a focus on the accuracy of results obtained via an integrated artificial intelligence (AI) tool. ROBOSPEC is a prospective, single-arm pilot study involving patients with prostate cancer undergoing robot-assisted radical prostatectomy (RARP). Probes from the RP specimens from the first 18 patients with intermediate-risk or high-risk prostate cancer were collected bilaterally from the dorsolateral sides of the prostate and examined with frozen section with hematoxylin and eosin staining (cryo-HE), SRH imaging (NIO laser imaging system, Invenio Imaging, Santa Clara, CA, USA). A previously published New York University AI algorithm (NYU-AI) that is based on the Inception-ResNet-v2 CNN architecture was used to generate three-color overlays to assist in interpretation. SRH images were reviewed by blinded urologists using this AI-enhanced output. NYU-AI identified positive surgical margins in 22% of patients, with no statistically significant difference in comparison to cryo-HE ( p > 0.05). Patient-based analysis yielded sensitivity and a negative predictive value (NPV) of 1.0, specificity of 0.93, and a positive predictive value of 0.75. Sample-based analysis showed similar performance, with specificity of 0.97 and identical sensitivity and NPV. These findings indicate strong diagnostic agreement between NYU-AI and conventional intraoperative pathology. Limitations of the study include the small patient cohort, the single-center design, previous training of the NYU-AI tool on prostate biopsy and periprostatic surgical-bed samples, and the lack of testing of interobserver agreement. Our preliminary findings support the potential of SRH with NYU-AI for intraoperative detection of positive surgical margins during RARP. Implementation of this technique should be further discussed after more studies have been conducted. We looked at an artificial intelligence program using a method called stimulated Raman histology to assess the cancer status of the cutting margin during robot-assisted surgery to remove the prostate. Our preliminary results show that this method could be an alternative to the current standard as it provides accurate and faster results.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.002 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.002 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.002 | 0.002 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".