An algorithm for using deep learning convolutional neural networks with three dimensional depth sensor imaging in scoliosis detection

Abstract

BACKGROUND CONTEXT: Timely intervention in growing individuals, such as brace treatment, relies on early detection of adolescent idiopathic scoliosis (AIS). To this end, several screening methods have been implemented. However, these methods have limitations in predicting the Cobb angle. PURPOSE: This study aimed to evaluate the performance of a three-dimensional depth sensor imaging system with a deep learning algorithm, in predicting the Cobb angle in AIS. STUDY DESIGN: Retrospective analysis of prospectively collected, consecutive, nonrandomized series of patients at five scoliosis centers in Japan. PATIENT SAMPLE: One hundred and-sixty human subjects suspected to have AIS were included. OUTCOME MEASURES: Patient demographics, radiographic measurements, and predicted Cobb angle derived from the deep learning algorithm were the outcome measures for this study. METHODS: One hundred and sixty data files were shuffled into five datasets with 32 data files at random (dataset 1, 2, 3, 4, and 5) and five-fold cross validation was performed. The relationships between the actual and predicted Cobb angles were calculated using Pearson's correlation coefficient analyses. The prediction performances of the network models were evaluated using mean absolute error and root mean square error between the actual and predicted Cobb angles. The shuffling into five datasets and five-fold cross validation was conducted ten times. There were no studyspecific biases related to conflicts of interest. RESULTS: The correlation between the actual and the mean predicted Cobb angles was 0.91. The mean absolute error and root mean square error were 4.0˚and 5.4˚, respectively. The accuracy of the mean predicted Cobb angle was 94% for identifying a Cobb angle of ≥10˚and 89% for that of ≥20˚. FDA device/drug status: Not approved (SCOLIOMAP).

An artificial intelligence powered platform for auto-analyses of spine alignment irrespective of image quality with prospective validation

“…Considering that spine alignment analysis is critical in the diagnosis and further management planning of scoliosis, assessment reliability and speed are important. Previously published studies have reported the use of machine learning models for measuring CAs [28][29][30] of AI techniques in determining spine shape, the datasets used to build the models were relatively small with debatable reliability. Instead of directly predicting CAs, some recent studies promoted the use of convolutional architectures to detect vertebral landmarks, followed by the calculation of CAs 12 , 14 , 31 ].…”

“…Considering that spine alignment analysis is critical in the diagnosis and further management planning of scoliosis, assessment reliability and speed are important. Previously published studies have reported the use of machine learning models for measuring CAs 28 , 29 , 30 . Although these studies have shown the potential value of AI techniques in determining spine shape, the datasets used to build the models were relatively small with debatable reliability.…”

A Survey of Methods and Technologies Used for Diagnosis of Scoliosis

“…Deep learning algorithms (DLAs) from CNNs, which have been applied to the detection of idiopathic scoliosis, were developed using 2D images [66] or Moiré topography [17,89,90]. Kokabu et al [91] modified their system [92] to predict the Cobb angle even more accurately, which they successfully presented in their current publication.…”

Reliability of automated topographic measurements for spine deformity

“…While clinicians tend to focus on curve magnitude and progression, patients and families are often concerned with thoracic prominence as well as shoulder, trunk, and waist-crease asymmetry [ 1 – 3 ]. Validated assessment tools and classification systems for scoliosis have been developed based on geometric radiographic measurements [ 4 – 6 ], but only recently have surface-topographic measures been recognized as important, objective measurements that may correlate closely with both patient self-image and radiographic measures of deformity [ 7 – 9 ].…”