Download This Paper
Preprints with The Lancet is a collaboration between The Lancet Group of journals and SSRN to facilitate the open sharing of preprints for early engagement, community comment, and collaboration. Preprints available here are not Lancet publications or necessarily under review with a Lancet journal. These preprints are early-stage research papers that have not been peer-reviewed. The usual SSRN checks and a Lancet-specific check for appropriateness and transparency have been applied. The findings should not be used for clinical or public health decision-making or presented without highlighting these facts. For more information, please see the FAQs.
Artificial Intelligence System Improved the Microwave Ablation Effect of Hepatocellular Carcinoma for Less-Experienced Doctors
35 Pages Posted: 23 Jun 2023
More...Abstract
Background: Complete eradication of hepatocellular carcinoma (HCC) through ablation is highly dependent on doctors’ experience and tumor location. Our purpose is to develop an artificial intelligence-assisted microwave ablation planning system (MWA-PS) to improve ablation treatment effect.
Methods: Pre-ablative image (contrast-enhanced ultrasound or computerized tomography/ magnetic resonance imaging), clinical information and ablation parameters were collected from one center and used to build MWA-PS by multilayer perceptron, XGBoosting, gate recurrent unit, and long short-term memory network (LSTM). HCCs from four centers were collected as prospective cohort (NCT 05270642) to compare the local tumor progression (LTP), technical effective, complication and mean ablation energy among experienced-doctor (ED), less-experienced-doctor (LED) and less-experienced-doctor with MWA-PS (MWA-PS) groups.
Findings: 875 HCCs were collected and divided into training (n=788) and validation (n=87) sets. LSTM showed best performance in training and validation sets, and was used to build MWA-PS. MWA-PS could provide ablation power, time, needle number, type, and distance for each ablation process. 388 HCCs were prospectively collected and randomly divided into ED (n=130), LED (n=128) and MWA-PS (n=130) groups. With 15.3 months median follow-up, MWA-PS group showed lower LTP than LED group (4.7%vs11.6%, p=0.041) and comparable LTP to ED group (4.7%vs5.5%, p=0.770). For HCCs adjacent to capsule/vessels, MWA-PS group showed relatively lower LTP than LED group (6.3%vs20.7%, p=0.096) and even slight lower than ED group (6.3%vs9.7%, p=0.600). Three groups achieved comparable technical effective and complication rate, but mean ablation energy of MWA-PS group was lower than LED and ED groups (28±10kJ vs 38±26kJ, p<0.001; 28±10kJ vs 34±16kJ, p<0.001).
Interpretation: Artificial intelligence system can help less-experienced doctors achieve comparable local tumor-control effect eradication ability to experienced doctors, especially for HCCs adjacent to capsule/vessels.
Trial Registration: The study was registered at Clinical Trial (NCT05270642).
Funding: National Key Research and Development Program of China NO.2022YFC2405500; National Natural Science Foundation of China NO.92159305, 12126607, 82030047; PLA health special Foundation NO.20BJZ42.
Declaration of Interest: All authors disclosed no relevant relationships.
Ethical Approval: Approved by the ethics committees of four hospitals (hospitals are listed in supplementary material).
Keywords: Hepatocellular carcinoma, Microwave ablation, Artificial intelligence
Suggested Citation: Suggested Citation