Use of superpixels for improvement of inter-rater and intra-rater reliability during annotation of medical images.

In the context of automatic medical image segmentation based on statistical learning, raters' variability of ground truth segmentations in training datasets is a widely recognized issue. Indeed, the reference information is provided by experts but bias due to their knowledge may affect the quality of the ground truth data, thus hindering creation of robust and reliable datasets employed in segmentation, classification or detection tasks. In such a framework, automatic medical image segmentation would significantly benefit from utilizing some form of presegmentation during training data preparation process, which could lower the impact of experts' knowledge and reduce time-consuming labeling efforts. The present manuscript proposes a superpixels-driven procedure for annotating medical images. Three different superpixeling methods with two different number of superpixels were evaluated on three different medical segmentation tasks and compared with manual annotations. Within the superpixels-based annotation procedure medical experts interactively select superpixels of interest, apply manual corrections, when necessary, and then the accuracy of the annotations, the time needed to prepare them, and the number of manual corrections are assessed. In this study, it is proven that the proposed procedure reduces inter- and intra-rater variability leading to more reliable annotations datasets which, in turn, may be beneficial for the development of more robust classification or segmentation models. In addition, the proposed approach reduces time needed to prepare the annotations.

Fully automated 3D body composition analysis and its association with overall survival in head and neck squamous cell carcinoma patients.

Objectives: We developed a method for a fully automated deep-learning segmentation of tissues to investigate if 3D body composition measurements are significant for survival of Head and Neck Squamous Cell Carcinoma (HNSCC) patients. Methods: 3D segmentation of tissues including spine, spine muscles, abdominal muscles, subcutaneous adipose tissue (SAT), visceral adipose tissue (VAT), and internal organs within volumetric region limited by L1 and L5 levels was accomplished using deep convolutional segmentation architecture - U-net implemented in a nnUnet framework. It was trained on separate dataset of 560 single-channel CT slices and used for 3D segmentation of pre-radiotherapy (Pre-RT) and post-radiotherapy (Post-RT) whole body PET/CT or abdominal CT scans of 215 HNSCC patients. Percentages of tissues were used for overall survival analysis using Cox proportional hazard (PH) model. Results: Our deep learning model successfully segmented all mentioned tissues with Dice's coefficient exceeding 0.95. The 3D measurements including difference between Pre-RT and post-RT abdomen and spine muscles percentage, difference between Pre-RT and post-RT VAT percentage and sum of Pre-RT abdomen and spine muscles percentage together with BMI and Cancer Site were selected and significant at the level of 5% for the overall survival. Aside from Cancer Site, the lowest hazard ratio (HR) value (HR, 0.7527; 95% CI, 0.6487-0.8735; p = 0.000183) was observed for the difference between Pre-RT and post-RT abdomen and spine muscles percentage. Conclusion: Fully automated 3D quantitative measurements of body composition are significant for overall survival in Head and Neck Squamous Cell Carcinoma patients.

Deep Learning Algorithm for Differentiating Patients with a Healthy Liver from Patients with Liver Lesions Based on MR Images.

The problems in diagnosing the state of a vital organ such as the liver are complex and remain unresolved. These problems are underscored by frequently published studies on this issue. At the same time, demand for imaging diagnostics, preferably using a method that can detect the disease at the earliest possible stage, is constantly increasing. In this paper, we present liver diseases in the context of diagnosis, diagnostic problems, and possible elimination. We discuss the dataset and methods and present the stages of the pipeline we developed, leading to multiclass segmentation of the liver in multiparametric MR image into lesions and normal tissue. Finally, based on the processing results, each case is classified as either a healthy liver or a liver with lesions. For the training set, the AUC ROC is 0.925 (standard error 0.013 and a p-value less than 0.001), and for the test set, the AUC ROC is 0.852 (standard error 0.039 and a p-value less than 0.001). Further refinements to the proposed pipeline are also discussed. The proposed approach could be used in the detection of focal lesions in the liver and the description of liver tumors. Practical application of the developed multi-class segmentation method represents a key step toward standardizing the medical evaluation of focal lesions in the liver.

Benchmarking of Deep Architectures for Segmentation of Medical Images.

In recent years, there were many suggestions regarding modifications of the well-known U-Net architecture in order to improve its performance. The central motivation of this work is to provide a fair comparison of U-Net and its five extensions using identical conditions to disentangle the influence of model architecture, model training, and parameter settings on the performance of a trained model. For this purpose each of these six segmentation architectures is trained on the same nine data sets. The data sets are selected to cover various imaging modalities (X-rays, computed tomography, magnetic resonance imaging), single- and multi-class segmentation problems, and single- and multi-modal inputs. During the training, it is ensured that the data preprocessing, data set split into training, validation, and testing subsets, optimizer, learning rate change strategy, architecture depth, loss function, supervision and inference are exactly the same for all the architectures compared. Performance is evaluated in terms of Dice coefficient, surface Dice coefficient, average surface distance, Hausdorff distance, training, and prediction time. The main contribution of this experimental study is demonstrating that the architecture variants do not improve the quality of inference related to the basic U-Net architecture while resource demand rises.

Use of artificial intelligence in the imaging of sarcopenia: A narrative review of current status and perspectives.

Sarcopenia is a muscle disease which previously was associated only with aging, but in recent days it has been gaining more attention for its predictive value in a vast range of conditions and its potential link with overall health. Up to this point, evaluating sarcopenia with imaging methods has been time-consuming and dependent on the skills of the physician. The solution for this problem may be found in artificial intelligence, which may assist radiologists in repetitive tasks such as muscle segmentation and body-composition analysis. The major aim of this review was to find and present the current status and future perspectives of artificial intelligence in the imaging of sarcopenia. We searched the PubMed database to find articles concerning the use of artificial intelligence in diagnostic imaging and especially in body-composition analysis in the context of sarcopenia. We found that artificial-intelligence systems could potentially help with evaluating sarcopenia and better predicting outcomes in a vast range of clinical situations, which could get us closer to the true era of precision medicine.

Endovascular embolization of wide-necked bifurcation aneurysms with the use of pCONus device: A systematic review and meta-analysis.

BACKGROUND: The following study aimed to summarize the overall safety and efficacy of the pCONus device in the treatment of wide-necked bifurcation aneurysms. MATERIAL AND METHODS: Major electronic medical databases were thoroughly searched to identify relevant studies. Data regarding the type of included studies, type of aneurysm and its location, treatment results measured in Raymond-Roy Occlusion Class (RROC) and its complications, as well as patients' neurological outcome at the discharge were extracted from the eligible studies and included in the meta-analysis. The subgroup analyses dependent on the aneurysm rupture status were also conducted. RESULTS: A total of 8 studies (198 patients with 200 aneurysms) were included in this meta-analysis. The most common localization of the treated aneurysms was the middle cerebral artery with its pooled prevalence estimate (PPE) of 44.5%. Immediately after the procedure, RROC I (complete obliteration of the aneurysm) was observed in 46.8%, RROC II in 32.9%, while RROC III (residual aneurysm) in 20.3% of the patients. In a short-term follow-up, PPE of RROC I was 55.0%, RROC II 29.0%, and RROC III 16.1%. The PPE of intraprocedural complications was 17.3%, the most frequent were thromboembolic events, which were observed in 12.1% of all procedures. Poor neurological outcome at the discharge was noted with PPE of 9.6%. CONCLUSIONS: The effectiveness and safety of wide-necked bifurcation aneurysm treatment utilizing the pCONus device is moderate. There is still a need for a consensus regarding the effective antiplatelet regimen in furtherance of diminishing the rate of thromboembolic events.

The semi-automated algorithm for the detection of bone marrow oedema lesions in patients with axial spondyloarthritis.

The aim of the study was to create the efficient tool for semi-automated detection of bone marrow oedema lesions in patients with axial spondyloarthritis (axSpA). MRI examinations of 22 sacroiliac joints of patients with confirmed axSpA-related sacroiliitis (median SPARCC score: 14 points) were included into the study. Design of our algorithm is based on Maksymowych et al. evaluation method and consists of the following steps: manual segmentation of bones (T1W sequence), automated detection of reference signal region, sacroiliac joint central lines and ROIs, a division of ROIs into quadrants, automated detection of inflammatory changes (STIR sequence). As a gold standard, two sets of manual lesion delineations were created. Two approaches to the performance assessment of lesion detection were considered: pixel-wise (detections compared pixel by pixel) and quadrant-wise (quadrant to quadrant). Statistical analysis was performed using Spearman's correlation coefficient. Correlation coefficient obtained for pixel-wise comparison of semi-automated and manual detections was 0.87 (p = 0.001), while for quadrant-wise analysis was 0.83 (p = 0.001). The correlation between two sets of manual detections was 0.91 for pixel-wise comparison (p = 0.001) and 0.88 (p = 0.001) for quadrant-wise approach. Spearman's correlation between two manual assessments was not statistically different from the correlation between semi-automated and manual evaluations, both for pixel- (p = 0.14) and quadrant-wise (p = 0.17) analysis. Average single slice processing time: 0.64 ± 0.30 s. Our method allows for objective detection of bone marrow oedema lesions in patients with axSpA. The quantification of affected pixels and quadrants has comparable reliability to manual assessment.

Angiographic and clinical results of anterior communicating artery aneurysm endovascular treatment.

INTRODUCTION: The anterior communicating artery (ACoA) is the most common location of intracranial aneurysms, observed in 35% of cases. Endovascular treatment has become an alternative to surgical clipping and the primary method of choice. AIM: To assess the treatment results of ruptured and unruptured ACoA aneurysms and to assess the incidence of intraprocedural complications and various factors influencing these aspects. MATERIAL AND METHODS: One hundred and eleven embolizations of ACoA aneurysms (80.7% ruptured and 19.3% unruptured) were retrospectively analysed. The methods of embolization were: coiling, balloon-assisted coiling, stent-assisted coiling. Morphology and dimensions of aneurysms were assessed on 3D digital subtraction angiography (DSA) images. Medical records were analysed for patient's clinical status at admission, intraprocedural complications, follow-up examination and modified Rankin Scale (mRS) score 1 month after discharge. RESULTS: Immediately after the procedure 56.9% of patients had Raymond-Roy Occlusion Classification (RROC) class I, 37.6% class II and 5.5% class III. The overall intraprocedural complication rate was 6.6%. There were significantly more cases of bleeding (p = 0.012) and coil prolapse (p = 0.012) during the procedures ending with higher packing density. Twenty-eight (25.7%) patients died during hospital stay, 27 (96.4%) with ruptured aneurysm. In the follow-up of 41 patients, RROC was the same or improved in 73.2% of cases and recanalization occurred in 26.8%. Six patients with aneurysm recanalization underwent repeat embolization. CONCLUSIONS: Endovascular embolization of ACoA aneurysms is an effective and safe treatment method. The most powerful factor influencing the incidence of complications is packing density. Superior orientation of the dome, initial incomplete embolization and poor outcome in mRS scale are factors predisposing to ACoA aneurysm recurrence.

The usefulness of diffusion-weighted imaging (DWI) and dynamic contrast-enhanced (DCE) sequences visual assessment in the early diagnosis of axial spondyloarthritis.

The aim of the study was to compare the diagnostic efficacy of the visual assessment of diffusion-weighted imaging (DWI) and dynamic contrast-enhanced (DCE) sequences compared to the STIR sequence in the diagnostics of active sacroiliitis in the course of axial spondyloarthritis (axSpA). The study group consisted of 49 patients who had undergone multiparametric magnetic resonance imaging of the sacroiliac joints (SIJs) due to clinical suspicion of axSpA. Two independent observers retrospectively assessed four quadrants of the SIJs for the presence of subchondral bone marrow oedema/osteitis with the use of modified SPARCC score in sequences: STIR, DWI (with ADC map) and DCE. Diagnostic efficiency parameters were calculated for DWI and DCE sequence separately, using STIR sequence as a reference. Inter-observer agreement was evaluated with the use of κ coefficient. Patients' clinical symptoms were analysed to identify the group fulfilling the imaging arm of the ASAS criteria for axSpA. Overall, 46.9% (n = 23) of patients fulfilled the imaging arm of ASAS criteria for axial spondyloarthritis. DWI with ADC map: accuracy 95.6%, sensitivity 99.4%, specificity 54.0%. DCE sequence: accuracy 96.8%, sensitivity 98.4%, specificity 79.5%. The highest level of inter-observer agreement was achieved for STIR sequence (κ = 0.888), slightly lower for DCE sequence (κ = 0.773) and the lowest for DWI with ADC (κ = 0.674). Visual assessment of the DWI and DCE sequences has high accuracy and sensitivity of bone marrow oedema/osteitis detection, but the specificity and inter-observer agreement are poor, especially for the DWI sequence with ADC maps.

Cerebral Aneurysms: Are They Associated with Anatomic Variations of Carotid and Main Cerebral Arteries?

OBJECTIVE: Vascular anatomic variations are very common in the population, but their coexistence with life-threating vascular malformations remains unclear. The objective of the study was to assess the correlation between the presence of anatomic variants of both carotid and main cerebral arteries and the occurrence of cerebral aneurysms. METHODS: We analyzed examinations of 194 patients who underwent computed tomography angiography of the head and neck between September of 2016 and November of 2017. The assessed parameters were: the variation of the common carotid arteries origin, hypoplasia or duplication of any of the main cerebral arteries, carotid artery kinking, and the presence of cerebral aneurysms. RESULTS: We found 5.7% of patients had a cerebral aneurysm. Hypoplasia of at least 1 main cerebral artery occurred in 34.5% of patients, with the most frequent being the vertebral artery (21.6%). Duplication of at least 1 main cerebral artery was detected in 2.6%. Cerebral aneurysms occurred more often together with hypoplasia (P = 0.041; OR = 3.175) or duplication (P < 0.001; OR = 18.500) of at least 1 main cerebral artery. In 7.2% of cases, the origin of common carotid arteries created a true bovine arch, and in 4.1% the so-called bovine arch. Patients with alternative variants of aortic arch branching were more likely to have cerebral aneurysms (P = 0.002; OR = 5.903). We found 27.3% of patients had internal carotid arteries affected by kinking. Carotid kinking did not predispose the patient to the formation of a cerebral aneurysm (P = 0.378). CONCLUSIONS: There is an evident correspondence between the presence of vascular anatomic variants (abnormal origin of the carotid arteries, hypoplasia, or duplication of the main cerebral artery) and the occurrence of cerebral aneurysms.

Radiologic approach to axial spondyloarthritis: where are we now and where are we heading?

Current emphasis on diagnosing axial spondyloarthritis (axSpA) in early stage enforced the search for sensitive and specific diagnostic algorithms with the use of imaging methods. The aim of this review was to summarise current recommendations concerning the use of imaging techniques in diagnostics and monitoring of axSpA as well as to outline possible future directions of the development in this field. MEDLINE database was searched between March and April 2018. In the first phase, such keywords were applied: 'ASAS', 'EULAR', 'ASAS-EULAR', 'ASAS/OMERACT', 'axial spondyloarthritis', while in the second step: 'axial spondyloarthritis', 'ankylosing spondylitis', 'magnetic resonance imaging', 'computed tomography', and 'radiography', 'imaging'. An up-to-date summary of European League Against Rheumatism (EULAR) recommendations enriched with recent updates of Assessment of Spondyloarthritis International Society (ASAS) diagnostic criteria regarding imaging in axSpA course was created. Moreover, we outlined the role of new in this field, promising imaging techniques, such as diffusion-weighted imaging and dynamic contrast-enhanced sequences in magnetic resonance imaging (MRI) or low-dose computed tomography (CT). As precise monitoring of axSpA activity is vital, we reviewed the most precise methods: semiquantitative scores (e.g., Spondyloarthritis Research Consortium of Canada scores or CT Syndesmophyte Score) and quantitative analysis of MRI-based apparent diffusion coefficient or perfusion maps and enhancement curves. According to EULAR and ASAS recommendations, radiography and MRI still remain basic methods of axSpA diagnostics and monitoring. However, the knowledge of state-of-the-art international guidelines combined with the awareness of emerging imaging methods is the key to effective management of axSpA.

Telocytes in the female reproductive system: An overview of up-to-date knowledge.

Telocytes are emerging cell population localized in the stroma of numerous organs, characterized by a distinctive morphology - small cell body with very long, slender prolongations, termed telopodes. Those cells can be found in the whole female reproductive system: in the vagina, uterus, oviducts and ovaries, mammary glands and also in the placenta. In our review, we aim at complete and transparent revision of the current knowledge of telocytes' localization and function, enriched by the analysis of the possible future direction of development of their clinical applications. The function of telocytes in the reproductive system has not been fully elucidated yet; however, many researchers point at their role in the regulation of local microenvironment, myogenic contractile mechanism, bioelectrical signaling, immunomodulation and regulation of blood flow. Additionally, previous research suggests that telocytes might act as sex hormone level sensors and are connected with pregnancy maintenance. As the morphology and number of those cells change under pathological conditions, such as pre-eclampsia, endometriosis and ovarian failure, there is a chance that they may contribute to therapy of abovementioned conditions. The impact of telocytes on stem cells and angiogenesis has been proven in many organs, and may be useful in regenerative medicine of the female reproductive system. A recently found connection between the proliferation rate of breast cancer cells and stromal cells like telocytes might be a step forward to the management of mammary gland neoplasms.

Spleen enlargement assessment using computed tomography: which coefficient correlates the strongest with the real volume of the spleen?

PURPOSE: The aim of the study was to find which linear measurements, field and volume coefficients correlate the best with real volume of the spleen and can be further used for identification of splenomegaly. METHODS: Abdominal computed tomography (CT) examinations of 264 patients were retrospectively analyzed in terms of maximal length, maximal thickness, hilum thickness, maximal height, vertical height and estimated height. Spleen volume was manually measured in Vitrea software. Two- and three-dimensional coefficients were calculated through proper mathematical formulas from linear measurements. Splenomegaly cut-off: 314.5 cm3. Data were analyzed with use of Pearson correlation and χ2 test with statistical significance at p < 0.05. RESULTS: For single measurements, the correlation with real spleen volume was the strongest for maximal height (r = 0.804; p < 0.05). Among two-dimensional indexes, multiplication product of maximal length and vertical height reached the highest level of correlation with spleen volume (r = 0.923; p < 0.05) and had the highest sensitivity and specificity (94.3% and 93.0%, respectively) for splenomegaly detection (threshold 115 cm2). In case of three-dimensional ones, the coefficient calculated from maximal length, vertical height and hilum thickness established the strongest link with spleen volume (r = 0.956; p < 0.05). CONCLUSIONS: Coefficient calculated from maximal length, vertical height and hilum thickness correlates the strongest with spleen volume and can be utilized for monitoring of spleen volume instead of obsolete splenic index. The most suitable for quick splenomegaly screening is two-dimensional coefficient (maximal length × vertical height), with the cut-off 115 cm2.

Nasal Septal Deviation and Concha Bullosa - Do They Have an Impact on Maxillary Sinus Volumes and Prevalence of Maxillary Sinusitis?

BACKGROUND: The aim of the study was to assess if the presence of nasal septal deviation and concha bullosa is connected with the development of sinuses and the incidence of inflammation within them. MATERIAL/METHODS: We retrospectively analysed 214 patients who underwent paranasal sinus computed tomography. There were 125 females and 89 males, the mean age being 47.67±16.74 years (range 18-97). Exclusion criteria included: age under 18 years, prior sinonasal surgery and S-shaped septum. RESULTS: Mean volume of the right maxillary sinus was 17.794 cm3, while for the left one it was 17.713 cm3. Nasal septal deviation was found in 79.9% of computed tomography examinations and concha bullosa was observed in 42.1% of the patients' examinations. There was an association between the presence of unilateral or dominant concha bullosa and contralateral direction of septal deviation [right-sided (p=0.039), left-sided (p=0.003)]. There was higher incidence of bilateral maxillary sinusitis in patients with septal deviation (p=0.007). Bilateral concha bullosa did not influence the incidence of bilateral maxillary sinusitis (p=0.495). Neither septal deviation (right sided: p=0.962; left-sided: p=0.731), nor unilateral/dominant concha bullosa (right: p=0.512; left: p=0,430) affected the asymmetry in volumes of maxillary sinuses. Bilateral concha bullosa was connected with larger volume of maxillary sinuses (right sinus: p=0.005; left sinus: p=0.048). CONCLUSIONS: Nasal septal deviation, contrary to concha bullosa, has influence on the development of maxillary sinusitis. There is a connection between the presence of concha bullosa and direction of septal deviation. Only bilateral concha bullosa affects maxillary sinus volumes.