This prospective, single-center, randomized controlled trial aims to evaluate the clinical effectiveness of an Image-Guided Platform assisted by a deep learning model for sentinel lymph node (SLN) retrieval in patients with early gastric cancer undergoing laparoscopic-endoscopic cooperative surgery combined with sentinel lymph node navigation surgery (LECS-SNNS). A total of 36 eligible participants will be randomly assigned in a 1:1 ratio to either the Image-Guided Platform group or the Traditional Palpation group. Both groups will undergo standardized indocyanine green fluorescence-guided sentinel lymphatic basin dissection and initial retrieval of fluorescent sentinel lymph nodes. In the experimental group, the remaining sentinel lymphatic basin (SLB) tissue will be examined using the Image-Guided Platform, which integrates radiographic and optical imaging with a deep learning model to identify and localize suspected lymph nodes. In the control group, LN retrieval will be performed using traditional visual inspection and palpation. The primary objective is to determine whether Image-Guided Platform-assisted LN retrieval increases the total number of SLNs retrieved intraoperatively compared with Traditional Palpation. Secondary objectives include assessment of LN retrieval efficiency, pathological findings, additional LN identification during permanent pathological examination, concordance between intraoperative frozen-section and permanent pathological examination, and other procedural, pathological, technical, and clinical outcomes relevant to the feasibility and performance of the Image-Guided Platform and deep learning model.
Sentinel lymph node navigation surgery provides an individualized approach to lymph node assessment in patients with early gastric cancer. However, the identification and retrieval of sentinel lymph nodes (SLNs) from resected sentinel lymphatic basin (SLB) tissue remain largely dependent on visual inspection and manual palpation by experienced physicians. This process can be time-consuming and operator-dependent and may result in missed lymph nodes (LNs), particularly small or visually inconspicuous nodes. This study is a single-center, prospective, randomized, superiority-controlled trial designed to evaluate an Image-Guided Platform assisted by a deep learning model for intraoperative and postoperative SLN retrieval in patients with early gastric cancer undergoing laparoscopic-endoscopic cooperative surgery combined with sentinel lymph node navigation surgery (LECS-SNNS). Eligible participants will be randomly assigned in a 1:1 ratio to either the Image-Guided Platform group or the Traditional Palpation group. Randomization will be performed using a computer-generated random sequence with allocation concealment. All participants will undergo a standardized LECS-SNNS procedure. Indocyanine green (ICG) will be injected endoscopically around the gastric lesion for near-infrared fluorescence lymphatic mapping. Fluorescent SLBs will be identified and resected en bloc. Fluorescent SLNs identified by visual inspection and fluorescence imaging will initially be retrieved and designated as SBNs. In the Image-Guided Platform group, the remaining SLB tissue will be placed on the Image-Guided Platform for radiographic and optical imaging. The acquired images will be analyzed using a deep learning model to identify and localize suspected LNs. Physicians will subsequently retrieve suspected LNs according to the image-guided and model-generated localization results. After image-guided LN retrieval, supplementary visual inspection and manual palpation will be performed to identify any additional LNs when appropriate. In the Traditional Palpation group, the remaining SLB tissue will be systematically examined by experienced physicians using standardized dissection, visual inspection, and manual palpation to identify and retrieve suspected LNs without image-guided or deep learning-assisted localization. All retrieved SLNs will undergo intraoperative frozen-section pathological examination according to standardized institutional procedures. Permanent pathological examination will subsequently be performed to confirm the pathological findings and to identify any additional LNs that were not retrieved during the intraoperative LN retrieval process. This approach will allow assessment of retrieved LNs, additional LNs identified during permanent pathological examination, and the overall LN harvest from the SLB specimen. Data Collection and Assessments Study data will include a broad range of demographic, clinical, diagnostic, surgical, pathological, technical, imaging-related, and perioperative information relevant to early gastric cancer, sentinel lymph node navigation surgery, LN retrieval, and pathological assessment. Data collection may include baseline demographic and clinical characteristics; medical and treatment history; preoperative laboratory, endoscopic, imaging, and clinical staging information; tumor-related characteristics; intraoperative and surgical variables; fluorescence mapping and SLB characteristics; LN retrieval and specimen-handling information; Image-Guided Platform-related parameters; radiographic and optical image-related parameters; deep learning model-related parameters; intraoperative frozen-section findings; permanent pathological findings; LN-related characteristics; perioperative recovery and safety information; and other clinically relevant variables collected according to the study protocol and case report forms. Derived variables, composite variables, subgroup classifications, technical performance measures, image- or specimen-level parameters, and exploratory clinical or pathological variables may also be generated from the collected data according to prespecified or exploratory statistical analyses. The collected information will be used to compare the effectiveness, efficiency, and diagnostic performance of Image-Guided Platform-assisted LN retrieval and Traditional Palpation, and to explore factors potentially associated with LN retrieval, total LN harvest, pathological assessment, technical performance, and other relevant clinical outcomes.
Study Type
INTERVENTIONAL
Allocation
RANDOMIZED
Purpose
OTHER
Masking
DOUBLE
Enrollment
36
The Image-Guided Retrieval Specimen Platform is an ex vivo LN retrieval system integrating radiographic imaging, optical imaging, and deep learning-based image analysis. Resected SLB tissue is placed on the platform for image acquisition. The acquired images are analyzed using a deep learning model to identify and localize regions suspected of containing LNs. Physicians subsequently retrieve suspected LNs according to the image-guided and model-generated localization results. All retrieved specimens are submitted for standardized pathological examination.
The residual SLB tissue will be systematically examined by experienced physicians using standardized dissection, visual inspection, and manual palpation to identify and retrieve suspected LNs. No Image-Guided Platform or deep learning-assisted localization will be used.
Beijing Friendship Hospital, Capital Medical University
Beijing, China
RECRUITINGTotal Number of Sentinel Lymph Nodes Retrieved Intraoperatively From the Sentinel Lymphatic Basin Specimen
The primary outcome is the per-participant total number of SLNs retrieved from the resected SLB specimen and submitted for intraoperative frozen-section pathological examination. The total number is defined as the sum of fluorescent SLNs identified during initial fluorescence-guided retrieval (SBNs) and additional non-fluorescent LNs retrieved from the residual SLB tissue during subsequent LN retrieval (BNs). The total number of intraoperatively retrieved SLNs will be compared between the Image-Guided Platform group and the Traditional Palpation group.
Time frame: During the operation, before completion of intraoperative frozen-section pathological assessment
Number of Fluorescent Specimens Not Pathologically Confirmed as Lymph Nodes
The number of ICG-fluorescent specimens initially retrieved as fluorescent SLNs but not subsequently confirmed as actual LNs by pathological examination will be recorded for each participant and compared between the two groups.
Time frame: From the operation through completion of permanent pathological examination, within 30 days after surgery
Number of Occult Lymph Nodes Identified During Permanent Pathological Examination
The number of additional LNs identified during permanent pathological examination of the residual SLB tissue after completion of intraoperative LN retrieval will be recorded for each participant. These LNs correspond to occult lymph nodes (OBNs) that were not identified during intraoperative LN retrieval.
Time frame: From the operation through completion of permanent pathological examination, within 30 days after surgery
Total Lymph Node Harvest From the Sentinel Lymphatic Basin Specimen
The total LN harvest from the resected SLB specimen will be recorded for each participant. Total LN harvest is defined as the sum of LNs retrieved during intraoperative LN retrieval and additional LNs identified during permanent pathological examination.
Time frame: From the operation through completion of permanent pathological examination, within 30 days after surgery
Proportion of Participants With Occult Lymph Nodes Identified During Permanent Pathological Examination
The proportion of participants in whom at least one occult LN is identified during permanent pathological examination after completion of intraoperative LN retrieval will be assessed.
Time frame: From the operation through completion of permanent pathological examination, within 30 days after surgery
Lymph Node-Level Concordance Rate Between Intraoperative Frozen-Section and Permanent Pathological Examination
For SLNs examined by both methods, the concordance rate will be calculated as the number of SLNs with concordant pathological classifications between intraoperative frozen-section and permanent pathological examination divided by the total number of SLNs examined by both methods, expressed as a percentage.
Time frame: From the operation through completion of permanent pathological examination, within 30 days after surgery
Lymph Node Retrieval Time
The time required to complete LN retrieval from the residual SLB tissue will be recorded for each participant and reported in minutes.
Time frame: During LN retrieval on the day of surgery
Intraoperative Sentinel Lymph Node Retrieval and Frozen-Section Pathology Workflow Time
The total workflow time will be recorded for each participant. Timing will begin when the residual SLB tissue, following initial retrieval of fluorescent SLNs, is transferred for further LN retrieval and pathological processing and will end when the intraoperative frozen-section pathology report is issued. The outcome will be reported in minutes.
Time frame: During the operation on the day of surgery
Number of Additional Lymph Nodes Retrieved by Supplementary Manual Palpation After Image-Guided Retrieval
In the Image-Guided Platform group, the number of additional LNs retrieved by supplementary visual inspection and manual palpation after completion of image-guided LN retrieval will be recorded. This outcome reflects LNs that were not initially identified during image-guided retrieval but were subsequently detected by supplementary manual palpation.
Time frame: During LN retrieval on the day of surgery
Number of Lymph Nodes Retrieved With the Image-Guided Platform
In the Image-Guided Platform group, the number of suspected LNs identified using the platform and subsequently retrieved for pathological examination will be recorded.
Time frame: During LN retrieval on the day of surgery
Pathological Confirmation Rate of Suspected Lymph Nodes Retrieved From Residual Sentinel Lymphatic Basin Tissue
The pathological confirmation rate will be assessed separately according to the LN retrieval method used for the residual SLB tissue after initial fluorescent SLN retrieval. For each retrieval method, the pathological confirmation rate will be calculated as the number of retrieved specimens pathologically confirmed as actual LNs divided by the total number of specimens retrieved as suspected LNs, expressed as a percentage.
Time frame: From LN retrieval through completion of permanent pathological examination, within 30 days after surgery
Lymph Node Detection Rate of the Image-Guided Platform
The LN detection rate of the Image-Guided Platform will be calculated as the number of pathologically confirmed actual LNs correctly identified by the Image-Guided Platform divided by the total number of actual LNs ultimately identified from the residual SLB tissue examined by the platform, expressed as a percentage. The denominator will include pathologically confirmed LNs retrieved with the Image-Guided Platform, additional LNs retrieved by supplementary manual palpation, and additional LNs subsequently identified during permanent pathological examination of the residual SLB tissue.
Time frame: From specimen imaging through completion of permanent pathological examination, within 30 days after surgery
This platform is for informational purposes only and does not constitute medical advice. Always consult a qualified healthcare professional.