Search

Traditional convolutional neural network (CNN) methods rely on dense tensors, which makes them suboptimal for spatially sparse data. In this paper, we propose a CNN model based on sparse tensors for efficient processing of high-resolution shapes represented as binary voxel occupancy grids. In contrast to a dense CNN that takes the entire voxel grid as input, a sparse CNN processes only on the non-empty voxels, thus reducing the memory and computation overhead caused by the sparse input data. We evaluate our method on two clinically relevant skull reconstruction tasks: (1) given a defective skull, reconstruct the complete skull (i.e., skull shape completion), and (2) given a coarse skull, reconstruct a high-resolution skull with fine geometric details (shape super-resolution). Our method outperforms its dense CNN-based counterparts in the skull reconstruction task quantitatively and qualitatively, while requiring substantially less memory for training and inference. We observed that, on the 3D skull data, the overall memory consumption of the sparse CNN grows approximately linearly during inference with respect to the image resolutions. During training, the memory usage remains clearly below increases in image resolution-an [Formula: see text] increase in voxel number leads to less than [Formula: see text] increase in memory requirements. Our study demonstrates the effectiveness of using a sparse CNN for skull reconstruction tasks, and our findings can be applied to other spatially sparse problems. We prove this by additional experimental results on other sparse medical datasets, like the aorta and the heart. Project page at https://github.com/Jianningli/SparseCNN ., (© 2023. The Author(s).)

This contribution presents the automatic segmentation of the lower jawbone (mandible) in humans' computed tomography (CT) images with the support of trained deep learning networks. CT acquisitions from the mandible frequently include radiological artifacts e.g., from metal dental restorations, ostheosynthesis materials or include trauma related free pieces of bones with missing bone contour anatomy. As a result, manual outlining these slices to generate the ground truth for evaluating segmentation algorithms lead to massive uncertainties and results in significant interphysician disagreement. Simply excluding these slices is also not the option of choice, regarding the treatment outcome. Hence, we defined strict inclusion and exclusion criteria for our datasets to avoid subjectivity or occurring bias in the groundtruth creation. Amongst others, datasets must display a complete physiological mandible without teeth. According to these data selection criteria such images are difficult to find since they originate from the clinical routine and therefore need a medical indication (such as trauma or pathologic lesions) to be provided as CT data. Furthermore, to prove the adequateness of our ground-truth, clinical experts segmented all cases twice manually, showing the great qualitative and quantitative agreement between them. Our dataset collection and the corresponding ground truth is an absolute novelty and the first serious evaluation of segmentation algorithms for the mandible.

Introduction: Computer assisted technologies based on algorithmic software segmentation are an increasing topic of interest in complex surgical cases. However-due to functional instability, time consuming software processes, personnel resources or licensed-based financial costs many segmentation processes are often outsourced from clinical centers to third parties and the industry. Therefore, the aim of this trial was to assess the practical feasibility of an easy available, functional stable and licensed-free segmentation approach to be used in the clinical practice., Material and Methods: In this retrospective, randomized, controlled trail the accuracy and accordance of the open-source based segmentation algorithm GrowCut was assessed through the comparison to the manually generated ground truth of the same anatomy using 10 CT lower jaw data-sets from the clinical routine. Assessment parameters were the segmentation time, the volume, the voxel number, the Dice Score and the Hausdorff distance., Results: Overall semi-automatic GrowCut segmentation times were about one minute. Mean Dice Score values of over 85% and Hausdorff Distances below 33.5 voxel could be achieved between the algorithmic GrowCut-based segmentations and the manual generated ground truth schemes. Statistical differences between the assessment parameters were not significant (p<0.05) and correlation coefficients were close to the value one (r > 0.94) for any of the comparison made between the two groups., Discussion: Complete functional stable and time saving segmentations with high accuracy and high positive correlation could be performed by the presented interactive open-source based approach. In the cranio-maxillofacial complex the used method could represent an algorithmic alternative for image-based segmentation in the clinical practice for e.g. surgical treatment planning or visualization of postoperative results and offers several advantages. Due to an open-source basis the used method could be further developed by other groups or specialists. Systematic comparisons to other segmentation approaches or with a greater data amount are areas of future works.

Ultrasound (US) is the most commonly used liver imaging modality worldwide. Due to its low cost, it is increasingly used in the follow-up of cancer patients with metastases localized in the liver. In this contribution, we present the results of an interactive segmentation approach for liver metastases in US acquisitions. A (semi-) automatic segmentation is still very challenging because of the low image quality and the low contrast between the metastasis and the surrounding liver tissue. Thus, the state of the art in clinical practice is still manual measurement and outlining of the metastases in the US images. We tackle the problem by providing an interactive segmentation approach providing real-time feedback of the segmentation results. The approach has been evaluated with typical US acquisitions from the clinical routine, and the datasets consisted of pancreatic cancer metastases. Even for difficult cases, satisfying segmentations results could be achieved because of the interactive real-time behavior of the approach. In total, 40 clinical images have been evaluated with our method by comparing the results against manual ground truth segmentations. This evaluation yielded to an average Dice Score of 85% and an average Hausdorff Distance of 13 pixels.

In this contribution, we present a semi-automatic segmentation algorithm for radiofrequency ablation (RFA) zones via optimal s-t-cuts. Our interactive graph-based approach builds upon a polyhedron to construct the graph and was specifically designed for computed tomography (CT) acquisitions from patients that had RFA treatments of Hepatocellular Carcinomas (HCC). For evaluation, we used twelve post-interventional CT datasets from the clinical routine and as evaluation metric we utilized the Dice Similarity Coefficient (DSC), which is commonly accepted for judging computer aided medical segmentation tasks. Compared with pure manual slice-by-slice expert segmentations from interventional radiologists, we were able to achieve a DSC of about eighty percent, which is sufficient for our clinical needs. Moreover, our approach was able to handle images containing (DSC=75.9%) and not containing (78.1%) the RFA needles still in place. Additionally, we found no statistically significant difference (p<;0.423) between the segmentation results of the subgroups for a Mann-Whitney test. Finally, to the best of our knowledge, this is the first time a segmentation approach for CT scans including the RFA needles is reported and we show why another state-of-the-art segmentation method fails for these cases. Intraoperative scans including an RFA probe are very critical in the clinical practice and need a very careful segmentation and inspection to avoid under-treatment, which may result in tumor recurrence (up to 40%). If the decision can be made during the intervention, an additional ablation can be performed without removing the entire needle. This decreases the patient stress and associated risks and costs of a separate intervention at a later date. Ultimately, the segmented ablation zone containing the RFA needle can be used for a precise ablation simulation as the real needle position is known.