Medical expert and machine learning analysis of lumbar disc herniation based on magnetic resonance imaging

https://doi.org/10.1016/j.cmpb.2021.106498Get rights and content

Highlights

  • •

    Convolutional neural network is used in lumbar disc MRI diagnostics.

  • •

    The Al2O3 spots (calcified foci) and lumbar disc fiber signals can be identified.

  • •

    Single factor analysis of variance is used to identify the smallest calcified focus and lumbar disc mass focus.

  • •

    The human observation and deep learning analysis is not significantly different (P<0.05).

Abstract

Objective

Observation and statistical analysis was used to evaluate the ability of lumbar disc magnetic resonance imaging (MRI) to obtain the smallest size of Al2O3 spots (calcified foci) and lumbar disc fiber signals.

Methods

First, we perform image acquisition of the MRI, perform the statistical analysis using five different sizes of Al2O3 spots and lumbar disc fibers on the imaging plate (IP), use a molybdenum target MRI machine 26 kV, adjust the milliampere amount, select the appropriate image processing parameters, and obtain the experimental image of the density value (D=0.70±0.05), the 5-point judgment method is used to obtain the total score of 10 lines of signals composed of 5 signals and noise, and a group is computed using the statistical analysis that is built from human observation and machine prediction (based on machine learning), which are then compared. In particular, we implemented a convolutional neural network algorithm to evaluate the medical condition against human observers, so as to study the structure of the lumbar intervertebral disc. We compute the true positive probability P(S/s) and false positive probability P(S/n) values, draw ROC curve, and compute the judgment probability value of each signal Pdet. We then use SPSS 10.0 statistical single factor analysis of variance software to process the data, and obtain the smallest calcified focus and lumbar disc mass focus.

Results

Using probability statistical methods to obtain the data of the ROC curve and the average value of the judgment probability Pdet, among 5 different sizes Al2O3 spots (calcifications), 0.20mm Pdet= 0.6250minimum, 0.55mm Pdet = 0.9000 the largest, but the difference between 0.20mm and 0.25mm Pdet is not statistically significant, and the difference is statistically significant; among the five types of lumbar disc fibers (tumor foci) of different sizes, 0.45mm Pdet= 0.5313minimum, 1.00mm Pdet =0.8813 is the largest, while the difference between 0.45mm and 0.60mm is not statistically significant, and the difference between 0.45mm and other is statistically significant. We note that the human observation and machine learning prediction is not significantly different (P<0.05).

Conclusions

The computation of the ROC curve and that of the probability of judgment using the statistical analysis based on a deep learning platform is simple and fast, and approximates that of human observation. It is suitable for the evaluation of image quality control carried out by daily clinical work.

Introduction

The spine is the backbone of the human torso, and the lumbar spine is the hub of human strength, so the diseases of the lumbar spine also appear with everyone's daily life and work. Therefore, understanding the structure of the lumbar intervertebral disc and being able to understand the MRI images of the lumbar and intervertebral discs is the golden key to mastering your own health [1].

The lumbar intervertebral disc is a complete functional structural unit, consisting of the annulus fibrosus and nucleus pulposus. Due to the influence of physiological and pathological factors, lumbar intervertebral discs are prone to degenerative changes, of which the degeneration of the nucleus pulposus is more obvious than that of the annulus fibrosus, which is mainly manifested by the decrease of water content, the decrease of proteoglycan concentration and the change of composition The mechanical properties also changed accordingly, causing a series of corresponding clinical symptoms and corresponding manifestations in imaging [8]. How to achieve early diagnosis of disc degeneration is related to clinical diagnosis, treatment and prognosis.

Imaging diagnosis refers to the strategy of diagnosing whether a patient has a specific disease through imaging indicators and evaluating the severity and prognosis of the disease. With the development of medical imaging technology, imaging indicators occupy an increasingly important position in the practice of medical diagnosis and treatment, and are used for the screening and diagnosis of many diseases [2]. Disease diagnosis research based on imaging indicators has also become a hot spot in medical research and one of the main development directions of medical diagnosis research. At present, many researchers have not adopted systematic research methods to carry out research, and failed to translate clinical findings into high-quality research evidence. This paper aims to discuss the statistical problems in MRI diagnosis research, hoping to help the selection of statistical strategies in subsequent research [3].

In practice, lumbar disc MRI is an important method for early detection of lumbar disc mass. Commonly used screen film cartilage (SFM) equipment, the detector used is a high-definition screen system dedicated to lumbar disc imaging, and MRI soft radiography mammography (CRM) equipment, used. The detector is an imaging plate (IP). For the evaluation of the amount of information on microscopic lesions displayed by these two kinds of special imaging equipment for lumbar discs, although it is comprehensively evaluated by fuzzy mathematical evaluation [2], the main evaluation method is still receiver operating characteristic (ROC). The traditional (ROC) analysis and evaluation of image quality is complicated, the data processing is time-consuming, and it is not easy to promote. In recent years, statistical analysis has developed new statistical models [3], which is small in size, simple to compute using a given equation, easy and fast to apply, and can be used in daily clinical image quality evaluation. In this study, the statistical analysis was used to evaluate the imaging quality of the CRM system [4].

With the rapid development of telemedicine technology, the requirements for medical image processing are becoming higher. This paper gives a more comprehensive overview of magnetic resonance images based on statistical methods in the medical application field, and analyzes the characteristics and limitations of its methods [5]. The cross penetration of various disciplines is the future development direction of medical image acquisition technology. In the past, structural imaging diagnosis using MRI technology often relied on visual evaluation and a method of setting up areas of concern with high randomness, but now, computer diagnostic methods based on imaging analysis via machine learning and statistical analysis methods are increasingly popular. Based on the analysis of lumbar disc herniation [6] and associated problems with the spline, this paper will focus on the imaging aspects and statistical compilation that comes from the medical expert observations and machine prediction.

Section snippets

Magnetic resonance imaging of lumbar disc

MRI has gradually become the simplest and most accurate examination method for disc degeneration. However, conventional MRI scans cannot detect the early changes of the intervertebral disc in time, and their imaging findings are often inconsistent with the pain symptoms or the patient's grade. Relaxation time mapping (relaxation time mapping) imaging technology, as a noninvasive, sensitive and reliable imaging method, can reflect the changes of its internal molecular level in the early stage of

Experimental results

Statistical processing: statistical analysis software SPSS 10.0 version for statistical analysis and testing (See Table 1).

Experimental principle and data processing: There are 10 groups of lumbar disc MRI, and the size of each group of simulated bodies is different. Each group has 10 small cylindrical simulated bodies, 5 of which have a signal (s), which represents the lesion; in addition, 5 noises (n), representing no lesions, as shown in Fig. 2 and Fig. 3. The observer is a medical

Discussion

In our experiment, it was found that under the condition that the IP board obtained a low contrast signal on the molybdenum target MRI machine, that is, using a statistical analysis model, that is, when the exposure value was selected so that the image density value D=0.70±0.05, there are 5 different sizes of the phantom Al2O3 spots (calcified foci) and 5 different sizes of lumbar disc fibers (mass foci), so that in one image, you can detect calcified foci of different lumbar disc sizes, and

Conclusion

In summary, with the continuous development of medical technology such as deep learning of clinical scans, the pathological value of orthopedic image diagnosis is increasing. Using statistical analysis to make ROC curve and obtaining judgment probability value Pdet, the effect achieved in the experiment based on human and machine observers is excellent. The value of this statistical diagnosis model is not only to diagnose whether or not to suffer from this disease, its final value is to pass

Declaration of Competing Interest

The authors declare that they have no conflicts of interest.

Acknowledgements

This work is supported by the medical core talents advanced learning and training project (No. 2019GG007).

References (15)

There are more references available in the full text version of this article.

Cited by (5)

#

These authors contributed equally to the work and should be considered as co-first authors.

View full text
© 2021 Elsevier B.V. All rights reserved.