Attention fusion network with self‐supervised learning for staging of osteonecrosis of the femoral head (ONFH) using multiple MR protocols

Abstract

BackgroundOsteonecrosis of the femoral head (ONFH) is characterized as bone cell death in the hip joint, involving a severe pain in the groin. The staging of ONFH is commonly based on Magnetic resonance imaging and computed tomography (CT), which are important for establishing effective treatment plans. There have been some attempts to automate ONFH staging using deep learning, but few of them used only MR images.PurposeTo propose a deep learning model for MR‐only ONFH staging, which can reduce additional cost and radiation exposure from the acquisition of CT images.MethodsWe integrated information from the MR images of five different imaging protocols by a newly proposed attention fusion method, which was composed of intra‐modality attention and inter‐modality attention. In addition, a self‐supervised learning was used to learn deep representations from a large amount of paired MR‐CT dataset. The encoder part of the MR‐CT translation network was used as a pretraining network for the staging, which aimed to overcome the lack of annotated data for staging. Ablation studies were performed to investigate the contributions of each proposed method. The area under the receiver operating characteristic curve (AUROC) was used to evaluate the performance of the networks.ResultsOur model improved the performance of the four‐way classification of the association research circulation osseous (ARCO) stage using MR images of the multiple protocols by 6.8%p in AUROC over a plain VGG network. Each proposed method increased the performance by 4.7%p (self‐supervised learning) and 2.6%p (attention fusion) in AUROC, which was demonstrated by the ablation experiments.ConclusionsWe have shown the feasibility of the MR‐only ONFH staging by using self‐supervised learning and attention fusion. A large amount of paired MR‐CT data in hospitals can be used to further improve the performance of the staging, and the proposed method has potential to be used in the diagnosis of various diseases that require staging from multiple MR protocols.

Applications of Deep Learning Models on the Medical Images of Osteonecrosis of the Femoral Head (ONFH): A Comprehensive Review

“…One more study was added from grey literature search. Finally, sixteen papers (n = 16) were eligible for data synthesis [33][34][35][36][37][38][39][40][41][42][43][44][45][46][47][48].…”

“…The authors of 16 articles were from four countries: China [34,[39][40][41][43][44][45][46][47][48], South Korea [33,35,36], Greece [37,38] and the United States [42]. Among the studies, two research groups, one from Greece [37,38] and one from China [43][44][45] published more than one article.…”

“…As shown in Table 1 and Fig. 2, magnetic resonance (MR) imaging was the most common modality (n = 10) [35,[37][38][39][41][42][43][44][45][46], while some investigations used digital radiographs or X-rays (n = 5) [33,34,36,40,48]. One study utilized computed tomography (CT) [47].…”

“…Out of the 16 papers that met the eligibility criteria for this review, only one did not report the number of participants or hips [42]. The remaining 15 papers accounted for a total of 73,341 hips, among which 19,168 belonged to ONFH patients, 47,389 to normal participants, and 6784 to patients with other hip pathologies [33][34][35][36][37][38][39][40][41][43][44][45][46][47][48]. Although 14 papers provided patient-level information, gender information was disclosed in only eight of them (Table 2).…”

“…Additionally, Li, et al [40] stated that their dataset was obtained from the hospital's repository. Kim, et al [35] also incorporated additional data into their dataset, which was sourced from the GoldAtlas dataset [49]. The GoldAtlas project dataset, which is publicly accessible, comprises a collection of MR and CT images.…”