International Society for Minimally Invasive Cardiothoracic Surgery

ISMICS Home ISMICS Home Past & Future Meetings Past & Future Meetings
Facebook   Instagram   Twitter   YouTube

Back to 2024 Abstracts

Do We Need Advanced Lung Segmentation Software?
Kostantinos Poulikidis, Syed Shahzad Razi, Russell Seth Martins, M. Jawad Latif, Jeffrey Luo, Faiz Bhora
Hackensack Meridian Health, Edison, NJ, USA

BACKGROUND: Understanding lung segmental anatomy is vital to the successful performance of anatomic lung segmentectomy. Variations in bronchovascular segmental anatomy are often challenging to understand based on conventional two-dimensional (2D) computed tomography (CT) scans. We present our initial experience with an automated, high-precision three-dimensional (3D) lung segmental reconstruction software to assist in preoperative planning.
METHODS: We used lung segmental reconstruction software (REVORAS, Ziosoft, Inc., Tokyo, Japan) that provides automated reconstruction of 3D images from 2D non-contrast CT scan (1.25mm slice intervals) based on bronchovascular anatomy.
The software provides surgeon-oriented 3D views with the hilar structures divided (or preserved) and the respective intersegmental planes. This allows for semi-automated segmentectomy planning based on the location of the tumor and/or simulation for additional bronchovascular division. The software also computes resection volume, and calculates surgical margins.
RESULTS: Table 1 summarizes important information related to the patients and operations, and highlights specific examples of the added utility provided by the three-dimensional (3D) lung segmental reconstruction software.
CONCLUSIONS:Softwares enabling 3D reconstruction from conventional CTs allow for more accurate identification of the target lung segment(s), help identify clinically relevant aberrant bronchovascular anatomy, and allow planning and simulation for appropriate parenchymal margins. For tumors traversing major fissures, 3D lung segmentation helps plan required configuration of en bloc resections. As machine learning becomes more sophisticated, we anticipate software such as these playing an increasingly important role in the automation of significant portions of thoracic operations.

Table: Patient and Operation Characteristics
Patient #DemographicsFEV1 and DLCO (% of predicted)Tumor CharacteristicsProcedurePathologyUtility of the Advanced Lung Segmentation Software
Patient 163F; past-smokerFEV1: 71% and DLCO: 77%1.2 cm in right upper lobeRight Upper Lobe Posterior (S2) SegmentectomyAdenocarcinomaIdentification of ascending A2 in the major fissure
Patient 272F; active-smokerFEV1: 55% and DLCO: 55%2.2 cm in right upper lobe/right middle lobeRight Middle/En-bloc S3 anterior segmentectomy; Completion upperlobectomyAdenocarcinomaCentral location of the tumor mandated right upper lobe/right middle lobe bi-lobectomy to achieve appropriate oncologic margins
Patient 363M; Past-smokerFEV1: 83% and DLCO: 106%1.5 cm in left lower lobeLeft lower lobe Basilar SegmentectomySquamous Cell CarcinomaIdentification of A7/8 perfusing S8; Inadequate surgical margin necessitated at least bisegmentectomy (S8-9)
Patient 475F; non-smokerFEV1: 95% and DLCO: 109%1.8 cm in right upper lobeRight Upper Lobe Posterior (S2) SegmentectomyAdenocarcinomaIdentification of the ground glass opacity (GGO) in S2; Identification of recurrent A2 from truncus anterior and appropriate identification of ascending branch in the fissure as the ascending A3
Patient 567F; past-smokerFEV1: 74% and DLCO: 20%1.1 cm in left upper lobeLeft upper lobe wedge resection with fiducial markerAdenocarcinomaIdentification of bronchovascular divisions of the left S3 as well as the intersegmental planes and resection volume

Back to 2024 Abstracts