Improvement of automated image stitching system for DR X-ray images

Artificial Intelligence in Medical Imaging: Analyzing the Performance of ChatGPT and Microsoft Bing in Scoliosis Detection and Cobb Angle Assessment

Open-source artificial intelligence models (OSAIM) find free applications in various industries, including information technology and medicine. Their clinical potential, especially in supporting diagnosis and therapy, is the subject of increasingly intensive research. Due to the growing interest in artificial intelligence (AI) for diagnostic purposes, we conducted a study evaluating the capabilities of AI models, including ChatGPT and Microsoft Bing, in the diagnosis of single-curve scoliosis based on posturographic radiological images. Two independent neurosurgeons assessed the degree of spinal deformation, selecting 23 cases of severe single-curve scoliosis. Each posturographic image was separately implemented onto each of the mentioned platforms using a set of formulated questions, starting from 'What do you see in the image?' and ending with a request to determine the Cobb angle. In the responses, we focused on how these AI models identify and interpret spinal deformations and how accurately they recognize the direction and type of scoliosis as well as vertebral rotation. The Intraclass Correlation Coefficient (ICC) with a 'two-way' model was used to assess the consistency of Cobb angle measurements, and its confidence intervals were determined using the F test. Differences in Cobb angle measurements between human assessments and the AI ChatGPT model were analyzed using metrics such as RMSEA, MSE, MPE, MAE, RMSLE, and MAPE, allowing for a comprehensive assessment of AI model performance from various statistical perspectives. The ChatGPT model achieved 100% effectiveness in detecting scoliosis in X-ray images, while the Bing model did not detect any scoliosis. However, ChatGPT had limited effectiveness (43.5%) in assessing Cobb angles, showing significant inaccuracy and discrepancy compared to human assessments. This model also had limited accuracy in determining the direction of spinal curvature, classifying the type of scoliosis, and detecting vertebral rotation. Overall, although ChatGPT demonstrated potential in detecting scoliosis, its abilities in assessing Cobb angles and other parameters were limited and inconsistent with expert assessments. These results underscore the need for comprehensive improvement of AI algorithms, including broader training with diverse X-ray images and advanced image processing techniques, before they can be considered as auxiliary in diagnosing scoliosis by specialists.

Evaluating Scoliosis Severity Based on Posturographic X-ray Images Using a Contrastive Language-Image Pretraining Model

Assessing severe scoliosis requires the analysis of posturographic X-ray images. One way to analyse these images may involve the use of open-source artificial intelligence models (OSAIMs), such as the contrastive language-image pretraining (CLIP) system, which was designed to combine images with text. This study aims to determine whether the CLIP model can recognise visible severe scoliosis in posturographic X-ray images. This study used 23 posturographic images of patients diagnosed with severe scoliosis that were evaluated by two independent neurosurgery specialists. Subsequently, the X-ray images were input into the CLIP system, where they were subjected to a series of questions with varying levels of difficulty and comprehension. The predictions obtained using the CLIP models in the form of probabilities ranging from 0 to 1 were compared with the actual data. To evaluate the quality of image recognition, true positives, false negatives, and sensitivity were determined. The results of this study show that the CLIP system can perform a basic assessment of X-ray images showing visible severe scoliosis with a high level of sensitivity. It can be assumed that, in the future, OSAIMs dedicated to image analysis may become commonly used to assess X-ray images, including those of scoliosis.

A novel multi-view X-ray digital imaging stitching algorithm

BACKGROUND

In fan beam X-ray imaging applications, several X-ray images sometimes need to be stitched together into a panoramic image because of the size limitations of the detector.

OBJECTIVE

This study aims to propose a novel multi-view X-ray digital imaging stitching algorithm (MVS) based on the CdZnTe photon counting linear array detectors to solve the problem of fan beam X-ray stitching deformation.

METHODS

The panoramic image is generated in four steps including (1) multi-view projection data acquisition, (2) overlapping positioning, (3) weighted fusion and (4) projected pixel value calculation. Images of a globe and foot are scanned by fan beam X-rays and a CdZnTe detector. The proposed method is applied to stitch together the scanned images of the globe. Three other methods are also used for comparison. Finally, this MVS algorithm is also used in the stitching of scanned images of the foot.

RESULTS

Compared with the 50% stitching accuracy of other methods, the new MVS algorithm reached a stitching accuracy of 94.4%. Image distortion on the globe and feet is also eliminated and thus image quality is significantly improved.

CONCLUSIONS

This study proposes a new multi-view X-ray digital imaging stitching algorithm. Study results demonstrate the superiority of this new algorithm and its feasibility in practical applications.

A new automatic stitching method for full-length lower limb radiography

Full-length lower limb x-rays are used to diagnose and plan surgical procedures, such as Total Knee Arthroplasty (TKA) and High Tibial Osteotomy (HTO). Due to the size limitation of digital radiography (DR), panoramic x-ray images cannot be obtained in a single exposure, necessitating multiple exposures and image stitching. In favor of manually constructing full-length x-ray images, we propose a new feature-based automated method for stitching together x-ray images. This new method is based on Canny algorithm, which detects and aligns bone edges before fusing them using a Wavelet form domain. Twenty-eight sets of lower limb x-ray images obtained from our hospital have been stitched and evaluated. The hip, knee, and ankle (HKA) angle was computed in two different ways then compared to manually stitched x-ray images by an expert. The stitching time was only three seconds, and the P-value was P = 0.974, and an accuracy rate of 100% was found. This method demonstrated greater precision and speed than both manually stitched x-ray images and previously published methods.

Heterogeneous Stitching of X-ray Images According to Homographic Evaluation

The C-arm X-ray system is a common intraoperative imaging modality used to observe the state of a fractured bone in orthopedic surgery. Using C-arm, the bone fragments are aligned during surgery, and their lengths and angles with respect to the entire bone are measured to verify the fracture reduction. Since the field-of-view of the C-arm is too narrow to visualize the entire bone, a panoramic X-ray image is utilized to enlarge it by stitching multiple images. To achieve X-ray image stitching with feature detection, the extraction of accurate and densely matched features within the overlap region between images is imperative. However, since the features are highly affected by the properties and sizes of the overlap regions in consecutive X-ray images, the accuracy and density of matched features cannot be guaranteed. To solve this problem, a heterogeneous stitching of X-ray images was proposed. This heterogeneous stitching was completed according to the overlap region based on homographic evaluation. To acquire sufficiently matched features within the limited overlap region, integrated feature detection was used to estimate a homography. The homography was then evaluated to confirm its accuracy. When the estimated homography was incorrect, local regions around the matched feature were derived from integrated feature detection and substituted to re-estimate the homography. Successful X-ray image stitching of the C-arm was achieved by estimating the optimal homography for each image. Based on phantom and ex-vivo experiments using the proposed method, we confirmed a panoramic X-ray image construction that was robust compared to the conventional methods.

Case of the Missing Vertebra: A Report of a Radiographic Stitching Error in a Scoliosis Patient

CASE

A 14-year-old girl with adolescent idiopathic scoliosis underwent imaging in preparation for scoliosis surgery. Posteroanterior traction radiographs showed 4 lumbar vertebrae, while the standing film showed 5. Reconciliation with the component radiographs used for the traction showed the discrepancy was caused by a software error. She underwent surgical correction, and her recovery has been uncomplicated.

CONCLUSION

Image stitching errors can lead to false depiction of structural abnormalities. Radiology technicians and clinicians should be cautious when reviewing digitally stitched images. We recommend that technicians label stitched images and indicate the overlapping region to assist with radiographic assessment.

Research on Panoramic Stitching Algorithm of Lateral Cranial Sequence Images in Dental Multifunctional Cone Beam Computed Tomography

In the design of dental multifunctional Cone Beam Computed Tomography, the linear scanning strategy not only saves equipment cost, but also avoids the demand for patients to be repositioned when acquiring lateral cranial sequence images. In order to obtain panoramic images, we propose a local normalized cross-correlation stitching algorithm based on Gaussian Mixture Model. Firstly, the Block-Matching and 3D filtering algorithm is used to remove quantum and impulse noises according to the characteristics of X-ray images; Then, the segmentation of the irrelevant region and the extraction of the region of interest are performed by Gaussian Mixture Model; The locally normalized cross-relation is used to complete the registration with the multi-resolution strategy based on wavelet transform and Particle Swarm Optimization algorithm; Finally, image fusion is achieved by the weighted smoothing fusion algorithm. The experimental results show that the panoramic image obtained by this method has significant performance in both subjective vision and objective quality evaluation and can be applied to preoperative diagnosis of clinical dental deformity and postoperative effect evaluation.

Effect of distortion corrections on the tractography quality in spinal cord diffusion-weighted imaging

PURPOSE

To assess the impact of a different distortion correction (DC) method and patient geometry (sagittal balance) on the quality of spinal cord tractography rendering according to different tractography approaches.

METHODS

Forty-four adults free of spinal cord diseases underwent cervical diffusion-weighted imaging. The phase-encoding direction was head→foot. Sequence with opposed polarities (foot→head) was acquired to perform DC. Eddy-current, motion effects, and susceptibility artifact correction methods were used for DC, and two deterministic and one probabilistic tractography approaches were evaluated using MRtrix and DSI Studio tractography software. Fiber length and number of fibers were extracted to evaluate the quality of the tractography rendering. For each subject, cervical lordosis was measured to assess patient geometry. The angle between the main direction of the spinal cord and the orientation of the acquisition box were computed at each spine level to assess acquisition geometry and define an angle threshold for which a tractography of good quality is no longer possible.

RESULTS

There was a significant improvement in tractography quality after performing DC with susceptibility artifact correction using a deterministic approach based on tensor. Before DC, the angle threshold was defined at C6 (15.2°) compared with C7 (21.9°) after corrections, demonstrating the importance of spinal cord angulation for DC.

CONCLUSION

The impact of DC on tractography quality is greatly impacted by acquisition geometry. To obtain a good-quality tractography, we propose as a future perspective to adapt the acquisition geometry to that of the patient by automatically adjusting the acquisition box.

A Convolutional Neural Network Based Auto-Positioning Method For Dental Arch In Rotational Panoramic Radiography

Dental panoramic radiography (DPR), a widely used medical examination method, has its intrinsic weakness in high requirement to the positioning of patient. Although positioning devices like chin support can provide a relatively stable and guaranteed environment for exposure, problems including morphological differences of jaw between patients and their improper standing postures still put the reconstructed image at high risk of getting blurred, especially in the anterior segment of dental arch. This paper proposes a novel method based on convolutional neural network (CNN) to estimate the positioning error of patient's dental arch, and thereby reconstruct the panoramic image with the corrected dental curvature, so that the blur gets reduced. Experiment results demonstrate the method's effectiveness in providing reconstructed images of stable quality for further diagnosis.