TSThomas Schnelldorfer

Thomas Schnelldorfer

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Thomas Schnelldorfer has a total of 55 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Development of a Deep Learning System for Intraoperative Identification of Cancer Metastases

    by Thomas Schnelldorfer, Janil Castro, Atoussa Goldar-Najafi, Liping Liu

    Objective: The aim of this study was to develop and test a prototype of a deep learning surgical guidance system [computer-assisted staging laparoscopy (CASL)] that can intraoperative identify peritoneal surface metastases on routine laparoscopy images. Background: For a number of cancer patients, operative resection with curative intent can end up in early recurrence of the cancer. Surgeons misidentifying visible peritoneal surface metastases is likely a common reason. Methods: CASL was developed and tested using staging laparoscopy images recorded from 132 patients with histologically confirmed adenocarcinoma involving the gastrointestinal tract. The data included images depicting 4287 visible peritoneal surface lesions and 3650 image patches of 365 biopsied peritoneal surface lesions. The prototype’s diagnostic performance was compared with results from a national survey evaluating 111 oncologic surgeons in a simulated clinical environment. Results: In a simulated environment, surgeons’ accuracy in correctly recommending a biopsy for metastases while omitting a biopsy for benign lesions was only 52%. In this environment, the prototype of a deep learning surgical guidance system demonstrated improved performance in identifying peritoneal surface metastases compared to oncologic surgeons with an area under the receiver operating characteristic curve of 0.69 (oncologic surgeon) versus 0.78 (CASL) versus 0.79 (human-computer combined). A proposed model would have improved the identification of metastases by 5% while reducing the number of unnecessary biopsies by 28% compared with current standard practice. Conclusions: Our findings demonstrate a pathway for an artificial intelligence system for intraoperative identification of peritoneal surface metastases but still require additional development and future validation in a multi-institutional clinical setting.


  • Can Surgeons Reliably Identify Non-cirrhotic Liver Disease During Laparoscopic Bariatric Surgery?

    by Alexandra Wuopio, Brett M. Baker, Benjamin Koethe, Martin D. Goodman, Reuben Shin, Nikolay Bugaev, Dmitry Nepomnayshy, Woon Cho Kim, Thomas Schnelldorfer

    Introduction Identification of liver disease during bariatric operations is an important task given the patients risk for occult fatty liver disease. Surgeon’s accuracy of assessing for liver disease during an operation is poorly understood. The objective was to measure surgeons’ performance on intra-operative visual assessment of the liver in a simulated environment. Methods Liver images from 100 patients who underwent laparoscopic bariatric surgery and pre-operative ultrasound elastography between July 2020 and July 2021 were retrospectively evaluated. The perception of 15 surgeons regarding the degree of hepatic steatosis and fibrosis was collected in a simulated clinical environment by survey and compared to results determined by ultrasonographic exam. Results The surgeons’ ability to correctly identify the class of steatosis and fibrosis was poor (accuracy 61% and 59%, respectively) with a very weak correlation between the surgeon’s predicted class and its true class (r = 0.17 and r = 0.12, respectively). When liver disease was present, surgeons completely missed its presence in 26% and 51% of steatosis and fibrosis, respectively. Digital image processing demonstrated that surgeons subjectively classified steatosis based on the “yellowness” of the liver and fibrosis based on texture of the liver, despite neither correlating with the true degree of liver disease. Conclusion Laparoscopic visual assessment of the liver surface for identification of non-cirrhotic liver disease was found to be an inaccurate method during laparoscopic bariatric surgery. While validation studies are needed, the results suggest the clinical need for alternative approaches. Graphical Abstract


  • Open Access

    White paper: requirements for routine data recording in the operating room

    by Thomas Schnelldorfer, Andrew A. Gumbs, Jennifer Tolkoff, Sarah Choksi, Jessica Stockheim, Amin Madani, Carla M. Pugh, Takeaki Ishizawa, Stefanie Speidel, Lee L. Swanström, et al.

    This white paper documents the consensus opinion of the authors and Artificial Intelligence Surgery editorial board members regarding common requirements needed to implement routine recording of data in the operating room. The statements were agreed upon by all authors and they attempted to outline common barriers that need to be addressed when implementing such recordings.


  • Open Access

    Evaluation of a polarization-enhanced laparoscopy prototype for improved intra-operative visualization of peritoneal metastases

    by Thomas Schnelldorfer, Einstein Gnanatheepam, Robert Trout, Ahmed Gado, Joyce-Ellen Pelletier, Long T. Dinh, Martin Hunter, Irene Georgakoudi

    Despite careful staging, the accuracy for preoperative detection of small distant metastases remains poor, creating a clinical need for enhanced operative staging to detect occult peritoneal metastases. This study evaluates a polarization-enhanced laparoscopy (PEL) prototype and assesses its potential for label-free contrast enhancement of peritoneal metastases. This is a first-in-human feasibility study, including 10 adult patients who underwent standard staging laparoscopy (SSL) for gastrointestinal malignancy along with PEL. Image frames of all detectable peritoneal lesions underwent analysis. Using Monte Carlo simulations, contrast enhancement based on the color dependence of PEL (mPEL) was assessed. The prototype performed safely, yet with limitations in illumination, fogging of the distal window, and image co-registration. Sixty-five lesions (56 presumed benign and 9 presumed malignant) from 3 patients represented the study sample. While most lesions were visible under human examination of both SSL and PEL videos, more lesions were apparent using SSL. However, this was likely due to reduced illumination under PEL. When controlling for such effects through direct comparisons of integrated (WLL) vs differential (PEL) polarization laparoscopy images, we found that PEL imaging yielded an over twofold Weber contrast enhancement over WLL. Further, enhancements in the discrimination between malignant and benign lesions were achieved by exploiting the PEL color contrast to enhance sensitivity to tissue scattering, influenced primarily by collagen. In conclusion, PEL appears safe and easy to integrate into the operating room. When controlling for the degree of illumination, image analysis suggested a potential for mPEL to provide improved visualization of metastases.


  • Open Access

    Development of a Deep Learning System for Intra-Operative Identification of Cancer Metastases

    by Thomas Schnelldorfer, Janil Castro, Atoussa Goldar-Najafi, Liping Liu

    For several cancer patients, operative resection with curative intent can end up in early recurrence of the cancer. Current limitations in peri-operative cancer staging and especially intra-operative misidentification of visible metastases is likely the main reason leading to unnecessary operative interventions in the affected individuals. Here, we evaluate whether an artificial intelligence (AI) system can improve recognition of peritoneal surface metastases on routine staging laparoscopy images from patients with gastrointestinal malignancies. In a simulated setting evaluating biopsied peritoneal lesions, a prototype deep learning surgical guidance system outperformed oncologic surgeons in identifying peritoneal surface metastases. In this environment the developed AI model would have improved the identification of metastases by 5% while reducing the number of unnecessary biopsies by 28% compared to current standard practice. Evaluating non-biopsied peritoneal lesions, the findings support the possibility that the AI system could identify peritoneal surface metastases that were falsely deemed benign in clinical practice. Our findings demonstrate the technical feasibility of an AI system for intra-operative identification of peritoneal surface metastases, but require future assessment in a multi-institutional clinical setting.

  • Long-Wave Infrared Imaging for Intraoperative Cancer Detection—What is the True Temperature of a Cancer?

    by Stephanie Vaughn, Robin Ruthazer, Andrew Rosenblatt, Roger L Jenkins, Andrea P Sorcini, Thomas Schnelldorfer

    Background: During cancer operations, the cancer itself is often hard to delineate—buried beneath healthy tissue and lacking discernable differences from the surrounding healthy organ. Long-wave infrared, or thermal, imaging poses a unique solution to this problem, allowing for the real-time label-free visualization of temperature deviations within the depth of tissues. The current study evaluated this technology for intraoperative cancer detection. Methods: In this diagnostic study, patients with gastrointestinal, hepatobiliary, and renal cancers underwent long-wave infrared imaging of the malignancy during routine operations. Results: It was found that 74% were clearly identifiable as hypothermic anomalies. The average temperature difference was 2.4°C (range 0.7 to 5.0) relative to the surrounding tissue. Cancers as deep as 3.3 cm from the surgical surface were visualized. Yet, 79% of the images had clinically relevant false positive signals [median 3 per image (range 0 to 10)] establishing an accuracy of 47%. Analysis suggests that the degree of temperature difference was primarily determined by features within the cancer and not peritumoral changes in the surrounding tissue. Conclusion: These findings provide important information on the unexpected hypothermal properties of intra-abdominal cancers, directions for future use of intraoperative long-wave infrared imaging, and new knowledge about the in vivo thermal energy expenditure of cancers and peritumoral tissue.


  • What is the Duct of Luschka?—A Systematic Review

    by Thomas Schnelldorfer, Michael G. Sarr, David B. Adams


  • Management of Giant Hemangioma of the Liver: Resection versus Observation

    by Thomas Schnelldorfer, Adam L. Ware, Rory Smoot, Cathy D. Schleck, William S. Harmsen, David M. Nagorney

    Management of patients with giant hemangiomas of the liver encounters persistent controversy. Although recent case series suggest a low complication rate with nonoperative management, the classic paradigm of preventive operative resection remains. A retrospective cohort study was conducted of 492 patients with giant hepatic hemangioma (>4 cm in size) diagnosed between 1985 and 2005 at Mayo Clinic Rochester. Long-term outcomes were assessed by patient survey, with a follow-up of 11 ± 6.4 years. Of 492 patients, 289 responded to the survey. In the nonoperative group (n = 233), 20% had persistent or new onset of hemangioma-associated symptoms, including potentially life-threatening complications in 2%. In the operative group (n = 56), perioperative complications occurred in 14%, including potentially life-threatening complications in 7%. None of the operative patients had persistent or new onset of hemangioma-associated symptoms after resection of the dominant hemangioma. In group comparison, the rate of adverse events was similar (20% versus 14%; p = 0.45) with an overall low risk for potentially life-threatening complications (2% versus 7%; p = 0.07). Size of hemangiomas was not associated with adverse events in either group. Subjective health status and quality of life at follow-up were similar in both groups (p > 0.54). Clinical observation of patients with giant hemangioma of the liver has a similar rate of complications compared with operative management, but might prevent the need for invasive interventions in some patients. Clinical observation is preferred in most patients and operative treatment should be reserved for patients with severe symptoms or disease-associated complications.


  • Alessandro Codivilla and the First Pancreatoduodenectomy

    by Thomas Schnelldorfer, Michael G. Sarr

    Opinion from JAMA Surgery — Alessandro Codivilla and the First Pancreatoduodenectomy