Danail Stoyanov

Objective. Detecting abnormal tissue in minimally invasive surgery is challenging due to the lack of direct palpation for the surgeon, with visualisation often being limited to white light endoscope images of the tissue surface, providing no information about inner structures. Imaging skins are being explored as a novel method of surgical image guidance: they are flexible and stretchable scintillating films that can act as indirect x-ray detectors when placed directly on the organ surface. Approach. Tomographic imaging of a custom quality evaluation phantom was performed using the imaging skin system. Image quality metrics, including spatial resolution, contrast, and uniformity, were obtained and compared to a conventional flat-panel micro-CT system. Main results. The imaging skin scan achieved a spatial resolution of 3.11 lp mm − 1 at 10% contrast. While these results were worse than that of the micro-CT system, the imaging skin system still met the requirements for clinical CT, demonstrating its clinical suitability. Significance. The feasibility of 3D x-ray imaging using imaging skins directly placed onto the imaged object has been demonstrated, achieving high resolution, comparable to clinical CT. Future work will involve artefact corrections and more advanced reconstruction techniques for the intended application of imaging skins for real-time surgical guidance.

Magnetic resonance imaging (MRI)-guided interventions have advanced rapidly due to their superior soft-tissue contrast and real-time imaging capability. To enable actuation for robots and other medical devices used in tumour biopsy and ablation within magnetically sensitive environments, several MRI-compatible pneumatic motors have been developed using 3D printing. Although 3D printing enables the fabrication of pneumatic structures, maintaining airtightness and smooth piston motion within 3D-printed cylinders remains challenging due to tolerance variations and surface roughness. These issues lead to pressure losses, increased air consumption, and reduced torque output during motor operation. To address these challenges, this paper presents a compact, high-performance, easily manufacturable MR Safe pneumatic stepper motor that circumvents 3D-printing tolerance limitations by employing prefabricated cylinders. The motor combines a replaceable design with great airtightness and high torque performance. The proposed design achieves negligible air leakage of 0.047 mL/min measured via a designed water displacement method and a maximum output torque of 1.04 Nm at 3.7 revolutions per minute (rpm) under 6 bar, without an additional gearbox. The motor has a compact design, 48 × 48 × 36 mm, which enables the motor to be integrated into the medical devices within the narrow MRI scanning bore.

Rapid advancements in diffusion models have enabled synthesis of realistic and anonymized imagery in radiography. However, due to their complexity, these models typically require large training volumes, often exceeding 10,000 images. Pre-training on natural images can partly mitigate this issue, but often fails to generate anatomically accurate shapes due to the significant domain gap. This prohibits applications in specialized medical conditions with limited data. We propose AnatoDiff, a diffusion model synthesizing high-quality X-Ray images with accurate anatomical shapes using only 500 to 1,000 training samples. AnatoDiff incorporates a Shape Prototype Module and Anatomical Fidelity loss, allowing for smaller training volumes through targeted supervision. We extensively validate AnatoDiff across three open-source datasets from distinct anatomical regions: Neonatal Abdomen (1,000 images); Adult Chest (500 images); and Humerus (500 images). Results demonstrate significant benefits, with an average improvement of 14.9% in Fréchet Inception Distance, 9.7% in Improved Precision, and 2.3% in Improved Recall compared to state-of-the-art (SOTA) few-shot and data-limited natural image synthesis methods. Unlike other models, AnatoDiff consistently generates anatomically correct images with accurate shapes. Additionally, a ResNet-50 classifier trained on AnatoDiff-generated images shows a 2.1% to 5.3% increase in F1-score, compared to being trained on SOTA diffusion images, across 500 to 10,000 samples. A survey with 10 medical professionals reveals that images generated by AnatoDiff are challenging to distinguish from real ones, with a Matthews correlation coefficient of 0.277 and Fleiss' Kappa of 0.126, highlighting the effectiveness of AnatoDiff in generating high-quality, anatomically accurate radiographs. Our code is available at https://github.com/KawaiYung/AnatoDiff.

Purpose: Video question answering (VideoQA) in the surgical domain aims to enhance intraoperative understanding by enabling AI models to reason over temporally coherent events rather than isolated frames. Current approaches are limited to static image features, and available datasets often lack temporal annotations, ignoring the dynamics critical for accurate procedural interpretation. Methods: We propose SurgViVQA, a surgical VideoQA model that extends visual reasoning from static images to dynamic surgical scenes. It uses a Masked Video–Text Encoder to fuse video and question features, capturing temporal cues like motion and tool–tissue interactions, which a fine-tuned LLM then decodes into coherent answers. To evaluate its performance, we curate REAL-Colon-VQA, a colonoscopic video dataset including motion questions and diagnostic attributes, including out-of-template questions with rephrased or semantically altered formulations to evaluate model robustness. Results: Experimental validation on REAL-Colon-VQA and the public EndoVis18-VQA dataset shows that SurgViVQA outperforms existing image-based VQA benchmark models and remains competitive with a fine-tuned video VLM baseline. In particular, SurgViVQA improves over PitVQA by +9% on REAL-Colon-VQA and +9% on EndoVis18-VQA in Keyword Accuracy, while achieving the strongest overall lexical and semantic generation performance. A perturbation study on the questions further confirms improved generalizability and robustness to variations in question phrasing. Conclusion: SurgViVQA and the REAL-Colon-VQA dataset provide a framework for temporally aware understanding in surgical VideoQA, enabling AI models to interpret dynamic procedural contexts.

Purpose:: With the growing demands for diverse X-ray imaging applications, there is a need for the development of flexible X-ray detectors to enable conformity to curved surfaces, improving spatial resolution, reducing X-ray exposure, and extending the potential of X-ray imaging to fields such as minimally invasive surgery, which would not be achievable using conventional rigid flat-panel detectors. Methods:: In this paper, we fabricate imaging skins—flexible and stretchable Gd2O2S:Tb-elastomer scintillator films—and integrate them into an indirect X-ray imaging system. We characterise X-ray imaging performance and mechanical properties of the imaging skins, investigating the impact of varying film fabrication parameters such as thickness and concentration. Results:: The results demonstrate the clinical suitability of the imaging skin system, achieving 1.05 to 2.18 lp/mm at 10% contrast despite the low X-ray energy used, which is comparable to clinical CT. Additionally, the high stretchability (255%) from the elastomer component was maintained in imaging skins with up to 50% Gd2O2S:Tb loading. Conclusion:: This work provides a basis for the integration of imaging skins in clinical settings, where they have the potential to be used for real-time intraoperative image guidance. Future work will investigate dynamic and tomographic imaging methods, determine effective deployment strategies of the imaging skins onto organ surfaces, and investigate the effects of imaging skin deformation on image quality.