MP18-20: Development of a 3D Tumor Model Fusion Pipeline for Renal Mass Ablation: Outcomes and Insights from a High-Volume Center
Friday, May 3, 2024 3:30 PM to 5:30 PM · 2 hr. (US/Central)
302B
Abstract
Information
Full Abstract and Figures
Author Block
Kamil Malshy*, Borivoj Golijanin, Bryan Jay, Yuanqing Bao, Emily Barry, Scott Collins, Aaron Maxwell, Nathaniel Rex, Gregory Dubel, Dragan Golijanin, Sari Khaleel, Providence, RI
Introduction
While prostate biopsy has shifted towards MRI-Ultrasound fusion with pre-procedural 3D simulation and probe placement planning, renal mass ablation (RMA) remains a “cognitive” procedure with no pre-treatment modeling. We describe a 3D tumor modeling and fusion radiomics pipeline developed at our institution using publicly available software, and its outcomes over a 5-year period.
Methods
Our pipeline utilizes 3D Slicer (version 4.10.2; Kilkinis et al, 2014), a popular open source radiomics program used for tumor segmentation and 3D modeling in oncology. Semi-automatically segmented 3D models from patient CT scans are used for simulation and probe placement planning, including color-coding the “safety” of ablation zones, then fused with our CT-guided procedural images in real time (Fig. 1). We retrospectively reviewed records of patients who underwent RMA using our pipeline. Outcomes included procedure safety, complications, RMA success and oncological outcomes. Data analysis was performed using R 4.3 (R Core Team, 2023)
Results
331 RMAs were conducted between 1/2018 and 6/2023. Average lesion size was 24mm (SD=9mm). Tumor location was mostly upper pole (36.7%), followed by lower and middle poles (36.1%, 27.1%, respectively). Most (305; 92.1%) patients had microwave vs cryoablation (26; 7.9%). Biopsy was done concurrently with RMA in 255 cases (76.6%) of cases, with most being 172 RCC (67.4%). Median microwave energy received per lesion was 45KJ (IQR = 27-60). 6 patients had Clavien-Dindo Grade 3a complications within 30 days of the procedure; none had higher grade complications. At a median follow-up of 2.1 years, 22 (6.6%) had tumor recurrences, 16 (4.8%) needed a second intervention (5 radical nephrectomy, 1 partial nephrectomy, and 11 redo ablations). 6 (2%) patients developed metastatic disease. Overall and cancer-specific mortality rates were 10% and 1% respectively after (IQR=0.9-3.8).
Conclusions
Using publicly available software, we developed a 3D modeling, treatment simulation, and real-time fusion pipeline for RMA that has been implemented at our institution for 5 years with favorable outcomes. Future directions include utilizing radiomic features of treated lesions, which are readily generated by the pipeline, to predict treatment outcomes with RMA using AI tools.
Source Of Funding
None
Author Block
Kamil Malshy*, Borivoj Golijanin, Bryan Jay, Yuanqing Bao, Emily Barry, Scott Collins, Aaron Maxwell, Nathaniel Rex, Gregory Dubel, Dragan Golijanin, Sari Khaleel, Providence, RI
Introduction
While prostate biopsy has shifted towards MRI-Ultrasound fusion with pre-procedural 3D simulation and probe placement planning, renal mass ablation (RMA) remains a “cognitive” procedure with no pre-treatment modeling. We describe a 3D tumor modeling and fusion radiomics pipeline developed at our institution using publicly available software, and its outcomes over a 5-year period.
Methods
Our pipeline utilizes 3D Slicer (version 4.10.2; Kilkinis et al, 2014), a popular open source radiomics program used for tumor segmentation and 3D modeling in oncology. Semi-automatically segmented 3D models from patient CT scans are used for simulation and probe placement planning, including color-coding the “safety” of ablation zones, then fused with our CT-guided procedural images in real time (Fig. 1). We retrospectively reviewed records of patients who underwent RMA using our pipeline. Outcomes included procedure safety, complications, RMA success and oncological outcomes. Data analysis was performed using R 4.3 (R Core Team, 2023)
Results
331 RMAs were conducted between 1/2018 and 6/2023. Average lesion size was 24mm (SD=9mm). Tumor location was mostly upper pole (36.7%), followed by lower and middle poles (36.1%, 27.1%, respectively). Most (305; 92.1%) patients had microwave vs cryoablation (26; 7.9%). Biopsy was done concurrently with RMA in 255 cases (76.6%) of cases, with most being 172 RCC (67.4%). Median microwave energy received per lesion was 45KJ (IQR = 27-60). 6 patients had Clavien-Dindo Grade 3a complications within 30 days of the procedure; none had higher grade complications. At a median follow-up of 2.1 years, 22 (6.6%) had tumor recurrences, 16 (4.8%) needed a second intervention (5 radical nephrectomy, 1 partial nephrectomy, and 11 redo ablations). 6 (2%) patients developed metastatic disease. Overall and cancer-specific mortality rates were 10% and 1% respectively after (IQR=0.9-3.8).
Conclusions
Using publicly available software, we developed a 3D modeling, treatment simulation, and real-time fusion pipeline for RMA that has been implemented at our institution for 5 years with favorable outcomes. Future directions include utilizing radiomic features of treated lesions, which are readily generated by the pipeline, to predict treatment outcomes with RMA using AI tools.
Source Of Funding
None
Sessions
MP18: Imaging/Uroradiology I
302B