While 3D medical shape generative models such as diffusion models have shown promise in synthesizing diverse and anatomically plausible structures, the absence of ground truth makes quality evaluation challenging. Existing evaluation metrics commonly measure distributional distances between training and generated sets, while the medical field requires assessing quality at the individual level for each generated shape, which demands labor-intensive expert review. In this paper, we investigate the use of classical machine learning (ML) methods and PointNet as an alternative, interpretable approach for assessing the quality of generated liver shapes. We sample point clouds from the surfaces of the generated liver shapes, extract handcrafted geometric features, and train a group of supervised ML and PointNet models to classify liver shapes as good or bad. These trained models are then used as proxy discriminators to assess the quality of synthetic liver shapes produced by generative models. Our results show that ML-based shape classifiers provide not only interpretable feedback but also complementary insights compared to expert evaluation. This suggests that ML classifiers can serve as lightweight, task-relevant quality metrics in 3D organ shape generation, supporting more transparent and clinically aligned evaluation protocols in medical shape modeling.

Intraperitoneal aerosolized drug delivery (IPADD) has emerged as a promising local treatment approach for peritoneal metastasis (PM). However, its effectiveness is currently limited by uneven aerosol distribution and restricted tumor tissue penetration. The use of electrostatic precipitation (ESP) exerts an electric force, which may enhance aerosol distribution and tissue penetration. A methylene blue (MB) solution was delivered intraperitoneally as an aerosol during laparoscopy in female pigs. Animals were randomized to receive either IPADD alone or IPADD combined with ESP. Tissue samples were taken from the ventral and lateral abdominal wall and from the small bowel. Aerosol distribution was evaluated using semiquantitative scoring of MB staining. Tissue penetration was measured on cryosections. Systemic resorption was assessed via serial blood sampling. Compared to IPADD alone, ESP clearly improved MB distribution, specifically in regions away from the nebulizer, and appreciably enhanced the tissue penetration in abdominal wall and small bowel samples. Sequenced activation of the three electrodes produced the most pronounced effects. ESP did not increase systemic exposure, and no tissue damage was observed. In this large animal model, multi-electrode ESP safely enhanced aerosol distribution and tissue penetration during IPADD, showing potential to improve treatment in patients with PM.

Background : Medical education increasingly incorporates active learning strategies, with dissection courses remaining essential for anatomical understanding. However, traditional dissection often lacks structured feedback. This study investigated the effectiveness of a peer-rubric evaluation on learning and motivation outcomes in a second-year medical course on the gastro-intestinal system. Methods: Students were randomized into a rubric group (n=105), who received peer evaluation based on a five-criterion rubric across seven dissection sessions, and a control group (n=309). Academic performance was assessed using a pretest, a practical posttest, and a theoretical examination. Student and evaluator experiences with the rubric were collected through surveys. Group differences in posttest and theoretical examination scores were analyzed using an independent Student’s T-test. Pearson correlation analysis assessed the relationship between difference in rubric scores and academic performance. A linear mixed-model analysis was conducted to evaluate trends in rubric scores across sessions. Results: No statistically significant differences were found in posttest (rubric group mean 15.50 ± 2.30/20; control 15.44 ± 2.75/20, p=0.86) or theory exam scores (rubric 14.22 ± 2.75/20; control 13.87 ± 3.27/20, p=0.33). A weak positive correlation (r=0.20, p=0.04) was observed between improvement in rubric scores over time and posttest and theory performance in the rubric group. Linear mixed-model analysis showed a significant upward trend in rubric scores across sessions (p<0.001), indicating improved performance over time. Survey data revealed mixed perceptions regarding peer evaluation’s objectivity and learning impact. Some students valued its role in promoting engagement, while others were sceptical about fairness and influence on final scores. Evaluators found the rubric clarified expectations but noted issues with grading consistency and workload. Conclusions: Despite no significant effect on summative outcomes, peer-rubric evaluation appears to enhance learning progression during practical sessions. Refining rubric design, strengthening assessor training, and adjusting assessment weighting may improve its effectiveness in anatomy education.

Digital media have reshaped anatomy education through the integration of virtual three-dimensional models. Techniques such as surface scanning, enable the development of highly photorealistic anatomical models; however, evidence regarding whether increased realism enhances learning outcomes remains mixed. Progress in this research area is further impeded by inconsistent conceptualizations of "realism." The present study examined the effects of geometric detail and shading in virtual pelvic models on anatomy knowledge, cognitive load, and motivation. Participants were randomly allocated to one of two learning conditions: (i) models with detailed geometry and texture shading (DGTS) or (ii) models with simplified geometry and simplified shading (SGSS). Baseline assessments included prior anatomical knowledge, spatial ability, and autonomous motivation. Learning outcomes were evaluated using cadaveric pelvic bones and prosections, assessing multiple cognitive levels as defined by the Blooming Anatomy Tool. Cognitive load was measured using the Paas scale, while autonomous motivation was assessed with the Academic Motivation Scale (Vallerand). No significant differences in anatomy knowledge acquisition were observed between the two conditions. However, spatial ability significantly interacted with posttest performance (r = 0.26, p < 0.01), with a stronger relationship in the DGTS condition (r = 0.29, p < 0.05), suggesting a trend toward greater dependency on spatial ability when learning with the detailed model. However, this difference between groups did not reach statistical significance. No group differences emerged for cognitive load or motivation. These findings indicate that increased realism alone does not guarantee improved educational effectiveness and highlight the importance of considering individual cognitive differences in instructional design.

Introduction The addition of bevacizumab (BEV) to chemotherapy has shown survival benefits in metastatic colorectal cancer but may increase operative risks, potentially limiting applicability in cytoreductive surgery (CRS) for CRPM. We report the phase II BEV-IP trial investigating the feasibility, safety and potential efficacy of BEV with neoadjuvant chemotherapy (NACT) before CRS in CRPM patients. Methods In this multicenter single-arm trial, 60 CRPM patients were enrolled to receive BEV and standard NACT prior to CRS. The primary endpoint was the 90-day major surgical morbidity rate. Secondary endpoints included safety profile of perioperative BEV, progression-free survival (PFS), overall survival (OS) and identification of exploratory biomarkers predictive of BEV response. Results A total of 54 patients were eligible for analysis. The 90-day major morbidity and mortality rates were 29.4% (95% CI 18.7 – 43.0%) and 3.9% (95% CI 1.1 – 13.2%), respectively. With a median follow-up of 71.4 months, the median PFS was 16.1 months (95% CI 11.2 – 21.0) and the median OS was 40.6 months (95% CI 32.4 – 48.8). A multivariable Cox-regression model identified peritoneal carcinomatosis index, BRAF mutation, peritoneal HbEGF levels and nuclear HIF1A staining as independent predictors of OS. Notably, in contrast to published literature, nuclear HIF1A overexpression was associated with improved OS (HR 0.17, 95% CI: 0.04–0.65,p = 0.010). Conclusion BEV-containing NACT for CRPM is safe and feasible in patients undergoing CRS. Peritoneal hbEGF concentrations and tumoral nuclear HIF1A staining emerge as potential biomarkers predictive of response to BEV, warranting further investigation of strategies targeting angiogenesis in CRPM.

Aim: To evaluate deep learning models for anatomical structure and peritoneal metastasis (PM) detection and segmentation during staging laparoscopy (SL) using a phase-independent dataset, and to quantify how annotation strategy and spatial representation relate to predictive performance. Methods: A checklist covering 25 anatomical structures, one surgical instrument, and PM was defined. Detection models (YOLOv9, Co-DETR) and segmentation models (SegFormer, Mask2Former) were trained under two label configurations. Videos were split at the video level (60/20/20). To quantify annotation distribution and spatial representation, two class-level descriptors were derived from the training set: object count and area fraction (percentage of image area occupied by each class). Class-level associations between these descriptors and test-set performance [F1-score, Intersection over Union (IoU)] were evaluated using Spearman correlation. Results: Thirty SL videos yielded 2,309 annotated frames (1,304/433/572 for training/validation/testing). YOLOv9 reached mean mAP@50 of 0.52 and 0.61; Mask2Former achieved mean IoU of 0.51 and 0.61 and F1-scores of 0.65 and 0.73 for Sets A and B, respectively. Despite 4,094 annotations, PM remained difficult to segment (IoU 0.29-0.30; F1-score 0.45-0.46), due to low area fraction and high heterogeneity. For IoU, area fraction showed stronger correlations with performance than object count (rho up to 0.66 vs. 0.48). Similar differences were observed for F1-score. Conclusions: Anatomical detection and segmentation during SL are feasible but limited by small-target representation and heterogeneous intra-abdominal context. Spatial representation is more closely associated with segmentation performance than annotation frequency, supporting annotation strategies that address sparse pixel coverage in phase-independent intra-abdominal models.

AIMS:To characterize the pharmacokinetics (PK), metabolite kinetics and toxicity risk of nanoparticle albumin-bound paclitaxel (Nab-PTX) administered via pressurized intraperitoneal aerosol chemotherapy (PIPAC) in patients with peritoneal metastasis, using data from a Phase 1 trial. METHODS:Twenty patients received two to three Nab-PTX PIPACs at doses ranging from 35 to 140 mg/m2 with a four-week dosing interval. A population pharmacokinetic-pharmacodynamic model was developed describing plasma concentrations of paclitaxel and metabolites (6-hydroxy and 3-hydroxy paclitaxel) and neutrophil dynamics. Bioavailability was estimated based on published intravenous data. Logistic regression was used to relate paclitaxel exposure to hepatotoxicity events. RESULTS:A two-compartment model with zero-order input best describes paclitaxel PK, with an estimated absorption duration of 3.49 h. Bioavailability was estimated at 37.0%. The total apparent volume of distribution of the two-compartment system was 2299 L. The population apparent clearance was 87.3 L/h and was accompanied by moderate variability between subjects and occasions (23.9 and 29.5% CV, respectively). Rate constants of metabolite formation and elimination displayed more extensive variability (52.6%-95.1% CV). Body surface area was incorporated as a covariate on apparent clearance.. Paclitaxel maximum concentrations and total exposure were associated with bilirubin and ALT adverse events, respectively, but not with other hepatotoxicity events. Dosing simulations predicted a low risk (0.8%) of grade 4 neutropenia at the highest investigated dose, with neutrophil nadirs at day 10 and recovery within 3-4 weeks. CONCLUSIONS:The developed model justifies a 140 mg/m2 Phase 2 dose characterized by slow absorption and tolerable dose-related haematological and hepatic toxicity.

Background: Chest tube insertions (CTI) have a high complication rate, warranting a dedicated Simulation-Based Mastery Learning (SBML) curriculum to acquire technical skills. This randomized controlled trial compares residents' skills in CTI after completing a SBML curriculum with those enrolled in a traditional residency program. Methods: Junior residents were baseline tested on cognitive and technical skills (Thiel bodies) before randomization into an intervention and control group. The former deliberately trained CTI on a porcine rib model until passing a predefined pass/fail score and were then summatively tested on Thiel bodies. The latter had no additional training opportunities and was evaluated 3 months later. Results: Seventeen residents were recruited and randomized. Following the per-protocol principle, a significant interaction effect for Group x Procedure (F(1,14) = 6.2, p = 0.026) was observed. Between baseline and summative assessment, both the control group (28.0 +/- 8.2 vs. 43.6 +/- 8.1, p < 0.001) and the intervention group (33.2 +/- 7.7 vs. 57.6 +/- 5.7, p < 0.001) significantly increased their scores. The intervention group outperformed the control group at summative assessment (43.6 +/- 8.1 vs. 57.6 +/- 5.7, p < 0.001). All participants in the intervention group and one resident in the control group achieved the pass/fail score. Conclusion: This SBML curriculum enabled quicker and superior skill acquisition. Skills trained on a porcine model are transferred to the highly realistic Thiel bodies and reach expert level, potentially increasing resident skill in clinical practice.

Although ultrasound (US) appears to complement traditional anatomy teaching, limited objective data exist on its efficacy. Existing literature often relies on student perceptions rather than performance-based outcomes. Additionally, the role of spatial understanding (SU)-the ability to mentally manipulate and interpret 3D anatomical relationships-and cognitive load (CL)-the mental effort required to learn-remains underexplored in the context of US-based instruction. The study consisted of three parts, with assessments before and after the US session. Prior to the session, students completed two paper-based tests on SU and cardiovascular system (CVS) anatomy. During the session, cardiac anatomy was explored through an introduction to US physics, a practical demonstration, and hands-on practice. Post-session, SU and CVS knowledge were reassessed, and participants completed a CL Scale Questionnaire. Thirty-one students participated in the study. Pre- and post-testing of CVS anatomy knowledge showed a mean increase of 11.33% (p < 0.05), while participants' mean SU scores improved from 65.71% to 81.04% (p < 0.05). The highest student rating on the CL Scale was observed when measuring the germane load, specifically the item assessing perceived learning (8.55 ± 1.31), while the lowest rating was reported for measurement of extraneous load, particularly the item assessing distractions (1.23 ± 1.61). This study provided insightful reports on the efficacy of US on SU and CL in anatomy education, showing its potential to improve learning outcomes and prepare students for clinical practice.

Medical image segmentation in laparoscopic surgery is challenging due to the high cost of pixel-level annotations and the need for domain expertise. In this paper, we investigate how Active Learning (AL) strategies perform across four laparoscopic video datasets that vary in two critical characteristics: (1) redundancy vs. diversity in their frames, and (2) balanced vs. imbalanced distributions of anatomical objects. We compare purely uncertainty-based approaches, purely diversity-based approaches, and hybrid strategies that combine both. Experimental results demonstrate that datasets containing many consecutive, visually similar frames hamper uncertainty-based AL methods, as these methods repeatedly select near-duplicate samples. In contrast, diverse datasets, especially those with balanced object frequencies, enable uncertainty-based methods to excel by focusing on hard or rare samples. For datasets with imbalanced object distributions, hybrid approaches prove particularly beneficial by capturing both model-driven uncertainty and underrepresented objects. Overall, our findings indicate that the choice of AL strategy should be carefully aligned with the inherent structure of a dataset, and that combining diversity and uncertainty often yields the most robust performance across various laparoscopic surgery scenarios. Our code is publicly available on https://github.com/amiiiirrrr/AL4LS .

Peritoneal metastases (PM) remain a significant clinical challenge due to their resistance to systemic chemotherapy. Intraperitoneal aerosolized drug delivery (IPADD) has shown promise as a localized treatment option; however, its efficacy is often limited by inadequate aerosol droplet distribution and poor tissue penetration. In this study, we enhance IPADD using electrostatic precipitation (eIPADD) and evaluate its performance through computational simulations and in vitro/ex vivo experiments. A realistic 3D reconstruction of the human peritoneal cavity was used to simulate the effects of electrostatic fields on droplet distribution and deposition. The results demonstrated that multiple electrodes improved aerosol homogeneity and tissue penetration depth. These findings were validated in vitro, where eIPADD significantly increased droplet coverage and tissue penetration, particularly in anatomically challenging regions. Importantly, increasing the number of electrodes from one to three further enhanced droplet distribution uniformity and tissue penetration depth. While raising the electrical potential improved deposition in key areas, benefits plateaued beyond 10 kV, suggesting a threshold in efficacy. Nevertheless, optimizing voltage within this range, in conjunction with the increased electrode count, remains critical for achieving consistent drug delivery across the peritoneal surfaces and maximizing therapeutic outcomes. Our findings suggest that eIPADD has the potential to address the limitations of conventional IPADD, providing a more effective and uniform drug delivery method for treating PM.

While the availability of open 3D medical shape datasets is increasing, offering substantial benefits to the research community, we have found that many of these datasets are, unfortunately, disorganized and contain artifacts. These issues limit the development and training of robust models, particularly for accurate 3D reconstruction tasks. In this paper, we examine the current state of available 3D liver shape datasets and propose a solution using diffusion models combined with implicit neural representations (INRs) to augment and expand existing datasets. Our approach utilizes the generative capabilities of diffusion models to create realistic, diverse 3D liver shapes, capturing a wide range of anatomical variations and addressing the problem of data scarcity. Experimental results indicate that our method enhances dataset diversity, providing a scalable solution to improve the accuracy and reliability of 3D liver reconstruction and generation in medical applications. Finally, we suggest that diffusion models can also be applied to other downstream tasks in 3D medical imaging. Our code is available at https://github.com/Khoa-NT/hyperdiffusion_liver

Video object segmentation is an emerging technology that is well-suited for real-time surgical video segmentation, offering valuable clinical assistance in the operating room by ensuring consistent frame tracking. However, its adoption is limited by the need for manual intervention to select the tracked object, making it impractical in surgical settings. In this work, we tackle this challenge with an innovative solution: using previously annotated frames from other patients as the tracking frames. We find that this unconventional approach can match or even surpass the performance of using patients' own tracking frames, enabling more autonomous and efficient AI-assisted surgical workflows. Furthermore, we analyze the benefits and limitations of this approach, highlighting its potential to enhance segmentation accuracy while reducing the need for manual input. Our findings provide insights into key factors influencing performance, offering a foundation for future research on optimizing cross-patient frame selection for real-time surgical video analysis.

Anatomy learning has traditionally relied on drawings, plastic models, and cadaver dissections/prosections to help students understand the three-dimensional (3D) relationships within the human body. However, the landscape of anatomy education has been transformed with the introduction of digital media. In this light, the Open Anatomy Explorer (OPANEX) was developed. It includes two user interfaces (UI): one for students and one for administrators. The administrator UI offers features such as uploading and labelling of 3D models, and customizing 3D settings. Additionally, the OPANEX facilitates content sharing between institutes through its import-export functionality. To evaluate the integration of OPANEX within the existing array of learning resources, a survey was conducted as part of the osteology course at Ghent University, Belgium. The survey aimed to investigate the frequency of use of five learning resources, attitudes towards 3D environments, and the OPANEX user experience. Analysis revealed that the OPANEX was the most frequently used resource. Students' attitudes towards 3D learning environments further supported this preference. Feedback on the OPANEX user experience indicated various reasons for its popularity, including the quality of the models, regional annotations, and customized learning content. In conclusion, the outcomes underscore the educational value of the OPANEX, reflecting students' positive attitudes towards 3D environments in anatomy education.

This exploratory study investigates anatomy faculty perceptions on the incorporation of ultrasound (US) in anatomy education, focusing on spatial understanding (SU) and cognitive load (CL) implications. An online survey was administered anonymously to members of the International Federation of Associations of Anatomists. The survey gathered quantitative and qualitative data, utilizing Likert scale- and open-ended questions. Quantitative results indicated a positive faculty disposition toward US as an educational tool, noting its potential to enhance SU by visualizing anatomical structures dynamically. However, concerns about increased CL due to technical and operational challenges were also highlighted. Thematic analysis of open-ended responses revealed three primary themes: US as a means for improved anatomical visualization, its role in reducing CL when appropriately integrated, and the logistical challenges associated with incorporating it into curricula. These findings align with prior research on student perspectives of US but provide a novel contribution by capturing the views of anatomy faculty. By offering faculty-driven insights, this study extends the literature on US in anatomy education and highlights the critical need for structured faculty training programs. The study underscores the importance of developing targeted US training for faculty to minimize CL impacts on both educators and students, optimizing the technology's educational potential. These insights contribute to a broader understanding of US in anatomy education and emphasize the need for institutional support, standardized curricula, and cost-effective implementation strategies. This research informs evidence-based curriculum development and practical integration strategies, aiding anatomy educators in making informed decisions about incorporating US into their teaching practices.

Real-time video segmentation is a promising opportunity for AI-assisted surgery, offering intraoperative guidance by identifying tools and anatomical structures. Despite growing interest in surgical video segmentation, annotation protocols vary widely across datasets – some provide dense, frame-by-frame labels, while others rely on sparse annotations sampled at low frame rates such as 1 FPS. In this study, we investigate how such inconsistencies in annotation density and frame rate sampling influence the evaluation of zero-shot segmentation models, using SAM2 as a case study for cholecystectomy procedures. Surprisingly, we find that under conventional sparse evaluation settings, lower frame rates can appear to outperform higher ones due to a smoothing effect that conceals temporal inconsistencies. However, when assessed under real-time streaming conditions, higher frame rates yield superior segmentation stability, particularly for dynamic objects like surgical graspers. To understand how these differences align with human perception, we conducted a survey among surgeons, nurses, and machine learning engineers and found that participants consistently preferred high-FPS segmentation overlays, reinforcing the importance of evaluating every frame in real-time applications rather than relying on sparse sampling strategies. Our findings highlight the risk of evaluation bias that is introduced by inconsistent dataset protocols and bring attention to the need for temporally fair benchmarking in surgical video AI.

Intraperitoneal aerosolized drug delivery (IPADD) is a minimally invasive technique for treating peritoneal metastasis (PM), combining laparoscopy with locoregional chemotherapy delivery as an aerosol. This study investigated key operating and physical parameters affecting IPADD performance, focusing on aerosolization pressure, droplet size distribution (DSD), and spray cone angle using six commercialized nebulizers: Nebulizer 770-12 (REGER Medizintechnik), CapnoPen (TM)(CapnoPharm), HurriChem (TM) (ThermaSolutions), MCR-4 TOPOL (TM) (Skala), QuattroJet (TM) (REGER Medizintechnik), and MiniJet (REGER Medizintechnik) in a reconstructed peritoneal cavity model. Results indicated notable variations in nebulizer performance. Nebulizer 770-12, CapnoPen, and HurriChem showed similar technical characteristics and reliable performance. DSD analysis showed bimodal distributions, with MiniJet and HurriChem producing small droplets (10-20 mu m), while MCR-4 TOPOL generated larger droplets due to its larger orifice and distinctive design. Spray cone angle measurements demonstrated that higher flow rates slightly improved dispersion, with the MCR-4 TOPOL achieving the widest angles. Optimal flow rates for uniform spray patterns varied, with CapnoPen and HurriChem performing well at lower rates, while MCR-4 TOPOL and QuattroJet required higher flow rates (>1.0 mL/s). This comprehensive evaluation provides valuable insights to optimize nebulizer selection and aerosolized drug delivery for improved IPADD efficacy.

BACKGROUND:Staging laparoscopy (SL) is an essential procedure for peritoneal metastasis (PM) detection. Although surgeons are expected to differentiate between benign and malignant lesions intraoperatively, this task remains difficult and error-prone. The aim of this study was to develop a novel multimodal machine learning (MML) model to differentiate PM from benign lesions by integrating morphologic characteristics with intraoperative SL images. MATERIALS AND METHODS:Deep learning (DL) models were trained to classify peritoneal lesions in video frames of patients undergoing SL for suspected PM. Two expert surgeons blinded to the pathology results performed an objective morphologic evaluation of these lesions. Traditional machine learning (ML) models were trained to predict tumors based on their morphology. A combined MML model was developed by integrating the best-performing morphology- and image-based models. The MML model was evaluated using an independent test set, and its predictions were compared with those of 13 oncologic surgeons. RESULTS:The cohort included videos of 67 patients, with 453 consecutive biopsied lesions (benign: n = 197; malignant: n = 256). The MML model achieved an area under the curve (AUC) of 0.88 [95% confidence interval (CI), 0.77-0.96], outperforming the best image-based DL model [AUC = 0.72 (95% CI, 0.54-0.87)], the best morphology-based ML [AUC = 0.86 (95% CI, 0.71-0.95)], and the surgeons' predictions [AUC = 0.78 (95% CI, 0.53-1.00)]. CONCLUSIONS:A novel MML model combining the visual and morphologic characteristics of peritoneal lesions was developed and internally validated, demonstrating good discriminative power for classifying PM during SL. This model shows promise as an intraoperative decision-support tool for surgeons, enhancing PM recognition and potentially reducing unnecessary biopsies.

INTRODUCTION:Pressurized intraperitoneal aerosol chemotherapy-oxaliplatin (PIPAC-OX) induces direct DNA damage and immunogenic cell death in patients with gastric cancer peritoneal metastases (GCPM). Combining PIPAC-OX with immune checkpoint inhibition remains untested. We conducted a phase I first-in-human trial evaluating the safety and efficacy of PIPAC-OX combined with systemic nivolumab (NCT03172416). METHODS:Patients with GCPM who experienced disease progression on at least first-line systemic therapy were recruited across three centers in Singapore and Belgium. Patients received PIPAC-OX at 90 mg/m2 every 6 weeks and i.v. nivolumab 240 mg every 2 weeks. Translational studies were carried out on GCPM samples acquired during PIPAC-OX procedures. RESULTS:In total, 18 patients with GCPM were prospectively recruited. The PIPAC-OX and nivolumab combination was well tolerated with manageable treatment-related adverse events, although one patient suffered from grade 4 vomiting. At second and third PIPAC-OX, respectively, the median decrease in peritoneal cancer index (PCI) was -5 (interquartile range: -12 to +1) and -7 (interquartile range: -6 to -20) and peritoneal regression grade 1 or 2 was observed in 66.7% (6/9) and 100% (3/3). Translational analyses of 43 GCPM samples revealed enrichment of immune/stromal infiltration and inflammatory signatures in peritoneal tumors after PIPAC-OX and nivolumab. M2 macrophages were reduced in treated peritoneal tumor samples while memory CD4+, CD8+ central memory and naive CD8+ T-cells were increased. CONCLUSIONS:The first-in-human trial combining PIPAC-OX and nivolumab demonstrated safety and tolerability, coupled with enhanced T-cell infiltration within peritoneal tumors. This trial sets the stage for future combinations of systemic immunotherapy with locoregional intraperitoneal treatments.

Laparoscopic exploration of the abdominal cavity is routinely performed for the diagnosis, assessment, and staging of peritoneal metastasis (PM). Accurately measuring tumor size during this procedure is crucial for prognosis and treatment planning. As conventional approaches for tumor size measurement rely on subjective manual assessments during or after surgery, they stand to benefit from computer assistance. This study proposes a new method for measuring tumor size in laparoscopic monocular videos. Specifically, we introduce a novel mathematical equation that connects the intrinsic parameters of a monocular camera, the surface area of target and reference objects, and their distances to the camera. Furthermore, we combine this equation with an object segmentation model (Mask2Former) and a depth estimation model (MiDaS), creating an end-to-end framework that automates tumor size measurement in monocular laparoscopic videos. We evaluate the proposed method using a laparoscopy dataset comprising 18 videos depicting 76 tumor biopsies, with tumor size measured by surgeons who are experts in laparoscopic surgery. When estimating the size of the various tumors in this dataset, we obtain a Mean Absolute Error (MAE) of 2.44 mm ± 0.23 mm, demonstrating that the newly proposed method accurately predicts intraoperative tumor size. Our code and the evaluation dataset are publicly available on https://github.com/amiiiirrrr/TSEMLV .