Surgical Phase Recognition: From Public Datasets to Real-World Data
Automated recognition of surgical phases is a prerequisite for computer-assisted analysis of surgeries. The research on phase recognition has been mostly driven by publicly available datasets of laparoscopic cholecystectomy (Lap Chole) videos. Yet, videos observed in real-world settings might contai...
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2022-08-01
|
| Series: | Applied Sciences |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2076-3417/12/17/8746 |
| _version_ | 1797496263598407680 |
|---|---|
| author | Kadir Kirtac Nizamettin Aydin Joël L. Lavanchy Guido Beldi Marco Smit Michael S. Woods Florian Aspart |
| author_facet | Kadir Kirtac Nizamettin Aydin Joël L. Lavanchy Guido Beldi Marco Smit Michael S. Woods Florian Aspart |
| author_sort | Kadir Kirtac |
| collection | DOAJ |
| description | Automated recognition of surgical phases is a prerequisite for computer-assisted analysis of surgeries. The research on phase recognition has been mostly driven by publicly available datasets of laparoscopic cholecystectomy (Lap Chole) videos. Yet, videos observed in real-world settings might contain challenges, such as additional phases and longer videos, which may be missing in curated public datasets. In this work, we study (i) the possible data distribution discrepancy between videos observed in a given medical center and videos from existing public datasets, and (ii) the potential impact of this distribution difference on model development. To this end, we gathered a large, private dataset of 384 Lap Chole videos. Our dataset contained all videos, including emergency surgeries and teaching cases, recorded in a continuous time frame of five years. We observed strong differences between our dataset and the most commonly used public dataset for surgical phase recognition, Cholec80. For instance, our videos were much longer, included additional phases, and had more complex transitions between phases. We further trained and compared several state-of-the-art phase recognition models on our dataset. The models’ performances greatly varied across surgical phases and videos. In particular, our results highlighted the challenge of recognizing extremely under-represented phases (usually missing in public datasets); the major phases were recognized with at least 76 percent recall. Overall, our results highlighted the need to better understand the distribution of the video data phase recognition models are trained on. |
| first_indexed | 2024-03-10T03:02:09Z |
| format | Article |
| id | doaj.art-cd638bc34c2e4ce48acc163351c30dac |
| institution | Directory Open Access Journal |
| issn | 2076-3417 |
| language | English |
| last_indexed | 2024-03-10T03:02:09Z |
| publishDate | 2022-08-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Applied Sciences |
| spelling | doaj.art-cd638bc34c2e4ce48acc163351c30dac2023-11-23T12:46:08ZengMDPI AGApplied Sciences2076-34172022-08-011217874610.3390/app12178746Surgical Phase Recognition: From Public Datasets to Real-World DataKadir Kirtac0Nizamettin Aydin1Joël L. Lavanchy2Guido Beldi3Marco Smit4Michael S. Woods5Florian Aspart6Caresyntax GmbH, Komturstr. 18A, 12099 Berlin, GermanyComputer Engineering Department, Yildiz Technical University, Istanbul 34220, TurkeyDepartment of Visceral Surgery and Medicine, Inselspital, Bern University Hospital, University of Bern, 3010 Bern, SwitzerlandDepartment of Visceral Surgery and Medicine, Inselspital, Bern University Hospital, University of Bern, 3010 Bern, SwitzerlandCaresyntax GmbH, Komturstr. 18A, 12099 Berlin, GermanyCaresyntax GmbH, Komturstr. 18A, 12099 Berlin, GermanyCaresyntax GmbH, Komturstr. 18A, 12099 Berlin, GermanyAutomated recognition of surgical phases is a prerequisite for computer-assisted analysis of surgeries. The research on phase recognition has been mostly driven by publicly available datasets of laparoscopic cholecystectomy (Lap Chole) videos. Yet, videos observed in real-world settings might contain challenges, such as additional phases and longer videos, which may be missing in curated public datasets. In this work, we study (i) the possible data distribution discrepancy between videos observed in a given medical center and videos from existing public datasets, and (ii) the potential impact of this distribution difference on model development. To this end, we gathered a large, private dataset of 384 Lap Chole videos. Our dataset contained all videos, including emergency surgeries and teaching cases, recorded in a continuous time frame of five years. We observed strong differences between our dataset and the most commonly used public dataset for surgical phase recognition, Cholec80. For instance, our videos were much longer, included additional phases, and had more complex transitions between phases. We further trained and compared several state-of-the-art phase recognition models on our dataset. The models’ performances greatly varied across surgical phases and videos. In particular, our results highlighted the challenge of recognizing extremely under-represented phases (usually missing in public datasets); the major phases were recognized with at least 76 percent recall. Overall, our results highlighted the need to better understand the distribution of the video data phase recognition models are trained on.https://www.mdpi.com/2076-3417/12/17/8746Laparoscopic videoscholecystectomydeep learningconvolutional neural networkphase recognitionsurgical data science |
| spellingShingle | Kadir Kirtac Nizamettin Aydin Joël L. Lavanchy Guido Beldi Marco Smit Michael S. Woods Florian Aspart Surgical Phase Recognition: From Public Datasets to Real-World Data Applied Sciences Laparoscopic videos cholecystectomy deep learning convolutional neural network phase recognition surgical data science |
| title | Surgical Phase Recognition: From Public Datasets to Real-World Data |
| title_full | Surgical Phase Recognition: From Public Datasets to Real-World Data |
| title_fullStr | Surgical Phase Recognition: From Public Datasets to Real-World Data |
| title_full_unstemmed | Surgical Phase Recognition: From Public Datasets to Real-World Data |
| title_short | Surgical Phase Recognition: From Public Datasets to Real-World Data |
| title_sort | surgical phase recognition from public datasets to real world data |
| topic | Laparoscopic videos cholecystectomy deep learning convolutional neural network phase recognition surgical data science |
| url | https://www.mdpi.com/2076-3417/12/17/8746 |
| work_keys_str_mv | AT kadirkirtac surgicalphaserecognitionfrompublicdatasetstorealworlddata AT nizamettinaydin surgicalphaserecognitionfrompublicdatasetstorealworlddata AT joelllavanchy surgicalphaserecognitionfrompublicdatasetstorealworlddata AT guidobeldi surgicalphaserecognitionfrompublicdatasetstorealworlddata AT marcosmit surgicalphaserecognitionfrompublicdatasetstorealworlddata AT michaelswoods surgicalphaserecognitionfrompublicdatasetstorealworlddata AT florianaspart surgicalphaserecognitionfrompublicdatasetstorealworlddata |