Three artificial intelligence data challenges based on CT and MRI.

PURPOSE: The second edition of the artificial intelligence (AI) data challenge was organized by the French Society of Radiology with the aim to: (i), work on relevant public health issues; (ii), build large, multicentre, high quality databases; and (iii), include three-dimensional (3D) information and prognostic questions. MATERIALS AND METHODS: Relevant clinical questions were proposed by French subspecialty colleges of radiology. Their feasibility was assessed by experts in the field of AI. A dedicated platform was set up for inclusion centers to safely upload their anonymized examinations in compliance with general data protection regulation. The quality of the database was checked by experts weekly with annotations performed by radiologists. Multidisciplinary teams competed between September 11th and October 13th 2019. RESULTS: Three questions were selected using different imaging and evaluation modalities, including: pulmonary nodule detection and classification from 3D computed tomography (CT), prediction of expanded disability status scale in multiple sclerosis using 3D magnetic resonance imaging (MRI) and segmentation of muscular surface for sarcopenia estimation from two-dimensional CT. A total of 4347 examinations were gathered of which only 6% were excluded. Three independent databases from 24 individual centers were created. A total of 143 participants were split into 20 multidisciplinary teams. CONCLUSION: Three data challenges with over 1200 general data protection regulation compliant CT or MRI examinations each were organized. Future challenges should be made with more complex situations combining histopathological or genetic information to resemble real life situations faced by radiologists in routine practice.

Impact of the rise of artificial intelligence in radiology: What do radiologists think?

PURPOSE: The purpose of this study was to assess the perception, knowledge, wishes and expectations of a sample of French radiologists towards the rise of artificial intelligence (AI) in radiology. MATERIAL AND METHOD: A general data protection regulation-compliant electronic survey was sent by e-mail to the 617 radiologists registered in the French departments of Nord and Pas-de-Calais (93 radiology residents and 524 senior radiologists), from both public and private institutions. The survey included 42 questions focusing on AI in radiology, and data were collected between January 16th and January 31st, 2019. The answers were analyzed together by a senior radiologist and a radiology resident. RESULTS: A total of 70 radiology residents and 200 senior radiologists participated to the survey, which corresponded to a response rate of 43.8% (270/617). One hundred ninety-eight radiologists (198/270; 73.3%) estimated they had received insufficient previous information on AI. Two hundred and fifty-five respondents (255/270; 94.4%) would consider attending a generic continuous medical education in this field and 187 (187/270; 69.3%) a technically advanced training on AI. Two hundred and fourteen respondents (214/270; 79.3%) thought that AI will have a positive impact on their future practice. The highest expectations were the lowering of imaging-related medical errors (219/270; 81%), followed by the lowering of the interpretation time of each examination (201/270; 74.4%) and the increase in the time spent with patients (141/270; 52.2%). CONCLUSION: While respondents had the feeling of receiving insufficient previous information on AI, they are willing to improve their knowledge and technical skills on this field. They share an optimistic view and think that AI will have a positive impact on their future practice. A lower risk of imaging-related medical errors and an increase in the time spent with patients are among their main expectations.

Artificial intelligence to diagnose meniscus tears on MRI.

PURPOSE: The purpose of this study was to build and evaluate a high-performance algorithm to detect and characterize the presence of a meniscus tear on magnetic resonance imaging examination (MRI) of the knee. MATERIAL AND METHODS: An algorithm was trained on a dataset of 1123 MR images of the knee. We separated the main task into three sub-tasks: first to detect the position of both horns, second to detect the presence of a tear, and last to determine the orientation of the tear. An algorithm based on fast-region convolutional neural network (CNN) and faster-region CNN, was developed to classify the tasks. The algorithm was thus used on a test dataset composed of 700 images for external validation. The performance metric was based on area under the curve (AUC) analysis for each task and a final weighted AUC encompassing the three tasks was calculated. RESULTS: The use of our algorithm yielded an AUC of 0.92 for the detection of the position of the two meniscal horns, of 0.94 for the presence of a meniscal tear and of 083 for determining the orientation of the tear, resulting in a final weighted AUC of 0.90. CONCLUSION: We demonstrate that our algorithm based on fast-region CNN is able to detect meniscal tears and is a first step towards developing more end-to-end artificial intelligence-powered diagnostic tools.

Automatic knee meniscus tear detection and orientation classification with Mask-RCNN.

PURPOSE: This work presents our contribution to a data challenge organized by the French Radiology Society during the Journées Francophones de Radiologie in October 2018. This challenge consisted in classifying MR images of the knee with respect to the presence of tears in the knee menisci, on meniscal tear location, and meniscal tear orientation. MATERIALS AND METHODS: We trained a mask region-based convolutional neural network (R-CNN) to explicitly localize normal and torn menisci, made it more robust with ensemble aggregation, and cascaded it into a shallow ConvNet to classify the orientation of the tear. RESULTS: Our approach predicted accurately tears in the database provided for the challenge. This strategy yielded a weighted AUC score of 0.906 for all three tasks, ranking first in this challenge. CONCLUSION: The extension of the database or the use of 3D data could contribute to further improve the performances especially for non-typical cases of extensively damaged menisci or multiple tears.

Perfusion MR imaging at 3-Tesla: Can it predict tumor grade and histologic necrosis rate of musculoskeletal sarcoma?

PURPOSE: To identify quantitative perfusion parameters that are best associated with tumor grade and tumor necrosis at magnetic resonance (MR) imaging at 3-Tesla. METHODS: MR perfusion studies of 31 patients with a musculoskeletal sarcoma were retrospectively evaluated by two readers. There were 18 men and 13 women with a mean age of 34.9±24.4 (standard deviation [SD] years) (range: 6-87 years). All patients underwent carcinologic tumor resection less than 3 months after MR imaging. For all patients six perfusion parameters (three semi-quantitative and three permeability parameters) were analyzed. The percentage of tumor necrosis was estimated using MR imaging. Perfusion data were compared between groups of tumors with different grades and necrosis ratios. Interobserver variability was calculated using intraclass correlation coefficient (ICC). RESULTS: Interobserver variability among the perfusion parameters was good to excellent (ICC: 0.72-0.9). The area under the curve and maximum slope values showed a significant association with the degree of tumor necrosis (P=0.02-0.04). When tumors with low necrosis ratios were compared to those with high ratios the former parameter was 80% lower. In the same groups, the imaging necrosis index was 56.9-59.8% higher in patients with grade 2 necrosis (P=0.01). Extracellular space volume (Ve) was 31.4% to 55.8% lower in tumors with high grade while the backflow constant (Kep) was 33.6% to 40.1%% higher in tumors with high grade. CONCLUSION: Semi-quantitative MR perfusion parameters have an excellent reproducibility and are associated with the degree of histologic tumor necrosis in musculoskeletal sarcomas. The utility of permeability parameters for determining tumor grade needs further investigations.

Results and limitations of humeral head resurfacing: 105 cases at a mean follow-up of 5 years.

BACKGROUND: The objective of this study was to assess clinical and computed-tomography (CT) outcomes at least 2 years after humeral head resurfacing to treat concentric gleno-humeral osteoarthritis. HYPOTHESIS: Humeral head resurfacing provides similar outcomes to those achieved with stemmed humeral head implants. MATERIALS AND METHODS: This single-centre retrospective study included 40 Copeland™ and 65 Aequalis™ humeral resurfacing heads implanted between 2004 and 2012. Mean patient age at diagnosis was 64 years. The diagnoses were osteoarthritis with an intact (68%) or torn (21%) rotator cuff, avascular necrosis (5%), osteoarthritis complicating chronic instability (3%), post-traumatic osteoarthritis (2%), and chronic inflammatory joint disease (1%). Validated clinical scores, radiographs, and CT before surgery and at last follow-up were compared. RESULTS: During the mean follow-up of 56 months, complications occurred in 24 implants. Revision surgery with reverse shoulder replacement was required in 18 cases, after a mean of 43.6 months, to treat glenoid wear or a rotator cuff tear. At last follow-up, for the implants that did not require revision surgery, the mean Constant score was 64/100. The implants had a mean varus of 5° and mean retroversion of -13.3°. The mean increase in glenoid cavity depth was 2.4mm. Mean increases in medial and lateral humeral offset were 1.9mm and 2.7mm, respectively. Pre-operative factors significantly associated with failure were rotator cuff tear (P=0.017) and glenoid erosion (P=0.001). DISCUSSION: We found a high failure rate related to glenoid wear or progressive rotator-cuff impairment, although CT showed no evidence of implant malposition or overstuffing. Previous studies of stemmed humeral head implants showed better outcomes. Given the low medium-term prosthesis survival rate, we now reserve humeral head resurfacing for concentric osteoarthritis without glenoid erosions or rotator cuff damage. LEVEL OF EVIDENCE: IV, retrospective study.

Correlation between primary flat foot and lower extremity rotational misalignment in adults.

PURPOSE: The goal of this study was to identify rotational abnormalities of the lower limb in adult patients with primary symptomatic flatfoot. MATERIALS AND METHODS: From September 2009 to May 2012, 24 patients (12 women, 12 men; mean age: 40 years) were prospectively included in the study. Each patient underwent radiographs of the flat foot and weight-bearing upright EOS® examination. Three-dimensional reconstructions of the lower extremities were performed with derived measurements (length, hip and knee parameters, rotations). A total of 31 symptomatic primary flat feet (bilateral flat foot, n=7 patients; unilateral flat foot, n=17 patients) were studied and compared to 30 control subjects matched for age and gender. A comparison between the two groups was made with the Student t-test. RESULTS: No significant differences were found between patients and control subjects on the coronal and sagittal planes. Similarly, no significant differences were observed between the 2 groups for rotation of the lower limbs (femoral torsion, tibial torsion, tibiofemoral rotation). CONCLUSION: There are no rotational abnormalities of the lower extremities in adult patients with primary symptomatic flat foot.

An anatomical study of the indirect tendon of the rectus femoris using ultrasonography.

UNLABELLED: The aim of this study was to demonstrate that ultrasound can allow a precise assessment of the indirect tendon of the rectus femoris using a new lateral approach. METHODS AND MATERIALS: Four hips were dissected for the anatomical study of the proximal rectus femoris insertions. Under ultrasonographic guidance, spinal needles piercing the direct tendon were placed in the indirect tendon, following by dissection. Then, high-resolution ultrasound was performed in 20 volunteers with assessment of the indirect tendon of the rectus femoris. RESULTS: At dissection, the spinal needles were located in or immediately adjacent to the indirect tendon, thus confirming that it was correctly depicted by ultrasound. The indirect tendon could be identified in each cadaver and each volunteer with ultrasound. The optimal position of the probe to allow assessment of the indirect tendon could be defined. No significant changes in the appearance or thickness of the tendon could be observed. CONCLUSION: The results of our study showed that the indirect tendon of the rectus femoris muscle can be clearly depicted by sonography in healthy adult subjects. The potential applications of this new use of sonography must now be confirmed by clinical studies. KEY POINTS: • The anatomy of the proximal rectus femoris is reviewed • Until now, sonography was limited for assessing direct and conjoined tendons • The indirect tendon can be clearly depicted by sonography • A new lateral approach for studying the indirect tendon is described.

Assessment of normal values of fractional anisotropy and mean diffusivity of mobile lumbar spine roots by diffuson tensor MRI: comparison between 1.5 and 3T.

PURPOSE: To assess the normal values of fractional anisotropy (FA) and mean diffusivity (MD) of L4, L5 and S1 nerve roots using diffusion tensor imaging (DTI) in healthy volunteers. MATERIALS AND METHODS: 37 subjects without previous history of lumbalgia or radiculalgia were prospectively examined: 27 at 1.5T and 10 at 3T MRI. The protocol included standard anatomical sequences and a DTI acquisition. Nerve root fibers were semi automatically extracted from DTI tractography. FA and MD values were measured at 4 key portions along each L4, L5 and S1 nerve roots. RESULTS: At 1.5T MRI, FA and MD were 0.221 ± 0.011 and 460.9 ± 35.5 mm2.s-1 respectively; at 3T MRI, FA and MD were 0.216 ± 0.01 and 480.1 ± 36.1 mm2.s-1 respectively, which may be considered as normal values for mobile lumbar spine nerve roots, independently of intersomatic space level (p = 0.06) and nerve root portion (p = 0.08) or magnetic field (p = 0.06). CONCLUSION: Normal FA and MD values can be measured along lumbar mobile spine nerve roots in healthy subjects. These values were not dependent on intersomatic space level, side or anatomical portion of the nerve root or magnetic field.

Pisotriquetral joint disorders: an under-recognized cause of ulnar side wrist pain.

Pisotriquetral joint disorders are often under-recognized in routine clinical practice. They nevertheless represent a significant cause of ulnar side wrist pain. The aim of this article is to present the main disorders of this joint and discuss the different imaging modalities that can be useful for its assessment.

The hand in systemic diseases other than rheumatoid arthritis.

This review outlines the skin, vascular and musculoskeletal symptoms affecting the hand during systemic inflammatory diseases other than rheumatoid arthritis. Skin lesions are diagnosed clinically and their symptomatology is documented through an extensive series of photographs. These conditions may require specific care before a surgical procedure can be performed. Vascular lesions are also diagnosed clinically and their symptomatology is described in detail. It is important to recognize that acrocyanosis is always benign. The surgeon should be able to distinguish between primary, but benign Raynaud's disease and secondary Raynaud's syndrome, which has a high risk of finger necrosis. Current preventative and curative treatments for finger necrosis are described. The clinical, radiological, progressive and therapeutic features of musculoskeletal lesions are reviewed, namely those associated with psoriatic arthritis, systemic sclerosis and lupus.

Clinical relevance of diffusion tensor imaging parameters in lumbar disco-radicular conflict.

PURPOSE: To measure the fractional anisotropy (FA) and the mean diffusivity (MD) values of L4, L5 and S1 nerve roots using diffusion tensor imaging (DTI) and to correlate them with four different clinical patterns. PATIENTS AND METHODS: Fifty-six human participants were prospectively included and divided between four groups: healthy subjects, patients with clinical symptomatic nerve root pain with and without anatomical discoradicular conflict and patients with incidental anatomical discoradicular conflict seen on magnetic resonance imaging (MRI). MRI protocol included anatomical sequences (sagittal T1- and T2-weighted, axial T2-weighted) and a 25 directions DTI sequence. FA and MD values were measured in consensus by two readers and compared between the four groups. RESULTS: Mean FA and MD values were significantly different for patients with clinically symptomatic nerve root pain (n=27) both with (n=16) (FA=0.187±0.015; MD=510±40) and without (n=11) (FA=0.193±0.011; MD=490±30.5) anatomical discoradicular conflict compared to healthy subjects (n=29) (FA=0.221±0.011; MD=460.9±35.5) including 2 subjects with incidental anatomical discoradicular conflict (FA=0.211±0.013; MD=450.8±41.2) on MRI (P=0.003). CONCLUSION: Measurement of FA and MD values of L4, L5 and S1 nerve roots using DTI could be useful in lumbar nerve root pain assessment. Further studies with different image processing methods are needed.

Anatomic and CT scan assessment of Teres Minor: a new index of trophicity.

INTRODUCTION: The present study reports the development of a CT assessment protocol for Teres Minor (TM) trophicity. HYPOTHESIS: Quantitative reproductible Terres Minor assessment on CT estimates the influence of muscle trophicity on the clinical and radiological results of palliative treatment of irreparable rotator cuff tear. MATERIALS AND METHOD: An anatomic study of 30 cadaveric shoulders confirmed a constant anatomic relation between Terres Minor and the inferior pole of the glenoid cavity. This landmark was used to develop a novel CT assessment of TM trophicity. RESULTS: The CT assessment showed excellent inter- and intra-observer reproductibility. The protocol defines a trophicity index, T2/G (T2 being TM thickness on axial CT slice, and G the maximum glenoid cavity thickness on axial slice), enabling reproductible TM analysis on preoperative arthro-CT. CONCLUSION: The study validated the CT protocol, allowing application in pre- and postoperative assessment of irreparable rotator cuff tear. LEVEL OF EVIDENCE: Level IV. Retrospective study.

Imaging of the nerves of the knee region.

Peripheral neuropathies are a frequent, but often underdiagnosed, cause of pain and functional impairment. The clinical symptoms can be subtle, and other neurologic or non neurologic clinical entities are often evoked. MRI and ultrasonography are the imaging modalities of choice for depicting nerves and assessing neuropathies. Common neuropathies in the knee area involve the saphenous, the tibial, the common peroneal and the sural nerves. The most frequent mechanisms of nerve injury in this area are nerve entrapment and nerve stretching. A perfect knowledge of the normal imaging anatomy is essential for accurate assessment of neuropathies. In this article, we will review the anatomy of the nerves around the knee, and their normal and pathological appearance.

Shelf arthroplasties long-term outcome: influence of labral tears. A prospective study at a minimal 16 years' follows up.

INTRODUCTION: Osteoarthritis lesions extent and dysplasia severity (negative vertical center edge [CE] angle) are recognized as unfavorable criteria for the survival of shelf arthroplasties performed for correcting hip dysplasia. Labral tears have recently been described on dysplastic hips, indicating beginning osteoarthritis and worsening the risk of instability. HYPOTHESIS: The labral tears identified in the course of shelf arthroplasty procedures for correction of hip dysplasia carry a predictive value for the survival of this operation. OBJECTIVES: Evaluate this hypothesis at the intermediate term in a long-term prospective observational study. PATIENT AND METHODS: Eighteen adult patients (18 dysplastic hips) having undergone shelf arthroplasty were included consecutively in a continuous prospective study. At the time the shelf arthroplasty was performed, a hip arthroscopic exam was carried out to search for and resect a labral tear if necessary. Fifteen patients were reviewed with a minimum follow-up of 16 years. Two patients died and one patient was lost to follow-up. RESULTS: During arthroscopic exploration, 10 hips presented labral tears (55.6%). At a mean follow-up of 16.3 years (range, 16-18 years), eight hips underwent hip arthroplasty. Of these hips, only one did not present a labral tear. The seven other hips had a tear of the labrum (p<0.001). The overall survival rate was 41.3%; it was 83.3% for hips with no labral tear and 15.2% for hips with a lesion of the labrum (p=0.048). DISCUSSION AND CONCLUSION: Labral tears had a negative impact on the outcome of shelf arthroplasty for hip dysplasia. This lesion therefore warrants being sought using appropriate exploration techniques (MRI or CT-arthrography) before shelf arthroplasty surgery. The existence of a preoperative labral tear does not seem to cast doubt on shelf arthroplasty itself. However, it should be identified so as to set objectives and expectations: long-term survival is significantly lower in the presence of a labral tear. It seems preferable to repair this type of lesion with arthroscopic guidance during shelf arthroplasty to prevent a potential source of residual pain, keeping in mind that secondary resection will be more difficult after covering the lesion. LEVEL OF EVIDENCE: Level 3 prospective observational prognostic study.

[Spinal infections: typical and atypical imaging features]. / Infections rachidiennes: aspects typiques et atypiques.

While the imaging features of established spondylodiscitis are well known, other presentations may be misleading. The purpose of this article is to review the typical imaging features of spondylodiscitis as well as some more atypical and unusual imaging features (early spondylodiscitis, sceptic spondylitis, septic arthritis of facet joints, primary epidural abscess, unusual germs, neonate).

Tractography of peripheral nerves and skeletal muscles.

The assessment of human peripheral nerves and skeletal muscles by means of diffusion tensor imaging and tractograpy has been a recent area of research. These techniques have been successfully applied in both volunteers and patients, providing non-invasively, quantitative microstructural parameters (mainly mean fractional anisotropy and apparent diffusion coefficient) and offering a three-dimensional visualization tool of nerves and muscles fibers. DTI and tractography may reveal abnormalities that are beyond the resolution of conventional MR techniques and hence open the way to potential clinical applications. In this article, we will first summarize the current state of DTI and tractography in the evaluation of peripheral nerves and skeletal muscles as well as their potential future clinical applications. Then, we will address important technical considerations, which understanding is necessary to appropriately apply DTI and tractograhy, and in order to understand the current limitations of these innovative and promising techniques.