Texture analysis based on U-Net neural network for intracranial hemorrhage identification predicts early enlargement

doi:10.1016/j.cmpb.2021.106140

Computer Methods and Programs in Biomedicine

Volume 206, July 2021, 106140

https://doi.org/10.1016/j.cmpb.2021.106140 Get rights and content

Highlights

•
The texture analysis based on U-Net neural network is helpful to predict the early expansion of hypertensive cerebral hemorrhage.
•
The characteristic value based on gray level co-occurrence matrix and gray level run length matrix can be used as an independent predictor.
•
The texture features of MI showed the lowest misclassified rate for prediction early expansion of hypertensive cerebral hemorrhage.

Abstract

Background and objective

Early hemorrhage enlargement in hypertensive cerebral hemorrhage indicates a poor prognosis. This study aims to predict the early enlargement of cerebral hemorrhage through the intelligent texture analysis of cerebral hemorrhage after segmentation.

Methods

A total of 54 patients with hypertensive intracerebral hemorrhage were selected and divided into enlarged hematoma (enlarged group) and non-enlarged hematoma (negative group). The U-Net Neural network model and contour recognition were used to extract the brain parenchymal region, and Mazda texture analysis software was used to extract regional features. The texture features were reduced by Fisher coefficient (Fisher), classification error probability combined average correlation coefficients (POE + ACC), and mutual information (MI) to select the best feature parameters. B11 module was used to analyze the selected features. The misclassified rate of feature parameters screened by different dimensionality reduction methods was calculated.

Results

The neural network based on U-Net can accurately identify the lesion of cerebral hemorrhage. Among the 54 patients, 18 were in the enlarged group and 36 in the negative group. The parameters of gray level co-occurrence matrix and gray level run length matrix can be used to predict the enlargement of intracerebral hemorrhage. Among the features screened by Fisher, POE + ACC and MI, the texture features of MI showed the lowest misclassified rate, which was 0.

Conclusion

The texture analysis based on U-Net neural network is helpful to predict the early expansion of hypertensive cerebral hemorrhage, and the parameters of gray level co-occurrence matrix and gray level run length matrix under MI dimensionality reduction have the most excellent predictive value.

Introduction

In recent years, the incidence of cerebrovascular diseases has increased year by year, among which cerebral hemorrhage is one of the most common symptoms [[1], [2], [3]]. Cerebral hemorrhage condition is critical, mainly in the middle-aged and elderly population, with the characteristics of disability, high fatality rate, and poor prognosis, which poses a severe threat to patients' life and health [4]. The incidence of early enlargement of intracerebral hemorrhage is 33%, which often indicates poor prognosis, with a fatality rate of up to 30% [5]. Effective prediction of the early expansion of spontaneous intracerebral hemorrhage has always been a challenging clinical problem.

With the continuous development and progress of modern medical detection methods, computed tomography (CT) imaging has played an increasingly important role in examining and treating cerebral hemorrhage due to its fast, accurate, and non-invasive characteristics [6,7]. However, the wide application of CT imaging has brought burdens to doctors and convenience to patients. For the diagnosis of cerebral hemorrhage, it has not proposed a machine measurement and calculation model, only by doctors based on experience to do qualitative analysis. Analyzing and reading a large amount of two-dimensional CT image data will cost doctors a lot of time and energy, and doctors with different seniority have different diagnosis and treatment levels, leading to the possibility of missed diagnosis and misdiagnosis in artificial diagnosis [5].

As an effective means of big data processing, artificial intelligence, especially deep learning, has been widely used in medical imaging research. The relevant deep learning algorithm represented by convolutional neural network and adversarial generation network is widely used in medical image annotation, feature classification, diagnosis and prediction, lesion location, medical image segmentation, medical image synthesis, medical image super-resolution reconstruction, and other directions [[8], [9], [10]]. In recent years, scholars at home and abroad have proposed the idea that hematoma has density heterogeneity. Based on baseline CT, some semi-quantitative imaging predictors have been proposed, but all of them have limitations of varying degrees. Radioomics is a new imaging post-processing method that extracts many image features from the image images and converts the image images into high-resolution. It excavates spatial data through high-throughput quantitative analysis. At present, it is mainly used in the study of tumor heterogeneity [11,12]. This study attempts to adopt the radiology method, based on the accurate identification of cerebral hemorrhage lesions by the U-Net neural network model, to quantitatively study the heterogeneity of cerebral hematoma through texture analysis, aiming to provide a more objective and sensitive prediction method for clinical practice.

Section snippets

General data

From January 2020 to January 2021, 54 patients with hypertensive intracerebral hemorrhage admitted to the Second Affiliated Hospital of Anhui Medical University were selected, including 36 males and 18 females, aged from 15 to 83 years old. Inclusion criteria: (1) history of hypertension; (2) clinical symptoms and signs are consistent with spontaneous cerebral hemorrhage; (3) the first CT scan was performed within six h after onset, and the second CT scan was performed within the nest 48 h; (4)

Incidence of hematoma enlargement

In 54 patients with intracranial hemorrhage, 18 cases of hematoma increased significantly, the occurrence rate was 33.3%. There were 11 males and 7 females paients in the enlarged group (54.7±15.2 years). The remaining 36 patients had no obvious changes in hematoma. There were 25 males and 11 females in the negative group (62.5±17.1 years).

Comparison of general data between the two groups

Except that there were statistically significant differences in Glasgow score, time of first CT examination, and hematoma volume between the two groups (P

Discussion

Scholars have tried to find the risk factors for the early expansion of hypertensive cerebral hemorrhage from the first CT scan in recent years. For morphology in the first CT scan, the hematoma volume > 30 ml, the irregular shape, and uneven density (mixed density, black holes, whirlpool sign) may indicate the early expansion. However, sensitivity and specialty rate is low and have too much subjectivity. The prediction of "point" sign in CTA is more accurate. However, due to contrast agent

Conclusions

In conclusion, this study confirmed that the CT texture analysis method could predict the expansion of cerebral hemorrhage. The parameters of gray level co-occurrence matrix and gray level run matrix under Mi dimensionality reduction have the most significant predictive value. The shortcoming of this study is that the sample size is not large enough. In the future, the sample size will be gradually increased, and more methods will be tried to enrich the study. It is hoped that more forces will

Statements of ethical approval

The study was approved by the Ethics Committee of The Anhui Medical University. All the participants provided their written informed consent to participate in this study.

Declaration of Competing Interest

There are no conflicts of interest to disclose for publication of this paper.

Acknowledgments

The project was funded by the National Natural Science Foundation of China (Reference No: 81200895), and the Interdisciplinary program of Anhui Medical University.

References (20)

F. Al-Mufti, K. Amuluru, A. Changa, M. Lander, N. Patel, E. Wajswol, S. Al-Marsoummi, B. Alzubaidi, I.P. Singh, R....
Y.A. Cavayas, L. Del Sorbo, E. Fan, Intracranial hemorrhage in adults on ECMO, Perfusion, 33 (2018)...
J.A. Frontera, J.J. Lewin, 3rd, A.A. Rabinstein, I.P. Aisiku, A.W. Alexandrov, A.M. Cook, G.J. del Zoppo, M.A. Kumar,...
R.L. Macdonald, Management of intracranial hemorrhage in the anticoagulated patient, Neurosurg. Clin. N. Am., 29 (2018)...
E. Marcolini, C. Stretz, K.M. DeWitt, Intracranial hemorrhage and intracranial hypertension, Emerg. Med. Clin. North...
H.W. Goo, J.M. Goo, Dual-energy CT: new horizon in medical imaging, Korean J. Radiol., 18 (2017)...
P. Rajiah, S. Abbara, CT coronary imaging-a fast evolving world, QJM Monthly J. Assoc. Phys., 111 (2018)...
M.P. McBee, O.A. Awan, A.T. Colucci, C.W. Ghobadi, N. Kadom, A.P. Kansagra, S. Tridandapani, W.F. Auffermann, Deep...
E. Moen, D. Bannon, T. Kudo, W. Graf, M. Covert, D. Van Valen, Deep learning for cellular image analysis, Nat. Methods,...
G. Zaharchuk, E. Gong, M. Wintermark, D. Rubin, C.P. Langlotz, Deep learning in neuroradiology, AJNR. Am. J....

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

Cited by (22)

Dual path sub-type diagnostic model based on the anatomic morphological feature
2024, Biomedical Signal Processing and Control
Intracranial hematoma can compress the brain tissue, resulting in increased cranial pressure and a high fatality rate. Therefore, providing a rapid and accurate clinical diagnosis and treatment plan is a crucial step. Intracranial hematoma can be classified into clinical sub-types such as cerebral lobes, ventricular, basal ganglia, pons, and cerebellum hemorrhage based on the location of the hemorrhage. To accurately identify different sub-types and provide targeted treatment, this paper proposes a dual path sub-type diagnosis model based on the anatomic morphological feature. It extracts anatomic morphological features of hematoma based on clinical prior knowledge to achieve prior feature enhancement. Then, a dual-path input model is constructed, and the dual-path module is used to analyze and integrate various features of different sub-types of intracranial hematoma. The model is validated on the clinical data set of the Second Affiliated Hospital of Dalian Medical University, and the corresponding accuracy can reach 0.993. This model can automatically recognize the sub-type of the clinical intracranial hematoma cases and improve diagnostic efficiency. Besides, it further aid in image analysis and enable specific treatment planning.
A Radiomics Model Based on CT Images Combined with Multiple Machine Learning Models to Predict the Prognosis of Spontaneous Intracerebral Hemorrhage
2024, World Neurosurgery
We aimed to construct 3 predictive models, including a clinical model, a radiomics model, and a combined model, to forecast the discharge prognosis of patients with intracerebral hemorrhage on admission.
A retrospective study was conducted, involving a total of 161 patients with intracerebral hemorrhage (ICH). At a ratio of 7:3, 115 of these patients were assigned to the training cohort, and 46 of these patients were assigned to the validation cohort. To produce the radionics signature and pick the features to use in its construction, the least absolute shrinkage and selection operator (LASSO) regression was applied. Five machine models were applied, and the optimal model was chosen to construct the radionics model. A clinical model was constructed using univariate and stepwise analysis to identify independent risk variables for poor outcomes at discharge. A predictive combined model nomogram was generated by integrating the clinical model and radiomics model. The performance of the nomogram was assessed in the training cohort and validated in the validation cohort. Analyses of the receiver operating characteristic curve (ROC), the calibration curve, and the decision curve were performed to assess the performance of the combined model.
This study encompassed a cohort of 161 individuals diagnosed with intracerebral hemorrhage (ICH), consisting of 110 males and 51 females. Utilizing the modified Rankin Scale (mRS) at discharge, the analysis revealed that 89 patients (55.3%) had a good prognosis, while 72 patients (44.7%) had a poor prognosis. Only 8 out of 1130 radiomics features were selected and used in conjunction with the LR algorithm to develop the radiomics model. Sex, IVH, GCS score, and ICH volume were determined to be independent predictors of poor outcomes at the time of discharge. The AUC values of the combined model, radiomics model, and clinical model were 0.8583, 0.8364, and 0.7579 in the training cohort, and 0.9153, 0.8692, and 0.7114 in the validation cohort, respectively. The combined model nomogram exhibited good calibration and clinical benefit in both the training and validation cohorts. The decision curve analysis (DCA) displays that the combined model obtained the highest net benefit compared to the radiomics model and clinics model in the training cohort.
The combined model demonstrates reliability and efficacy in predicting the discharge prognosis of ICH, enabling physicians to perform individualized risk assessments, and make optimal choices about patients with ICH.
CHSNet: Automatic lesion segmentation network guided by CT image features for acute cerebral hemorrhage
2023, Computers in Biology and Medicine
Stroke is a cerebrovascular disease that can lead to severe sequelae such as hemiplegia and mental retardation with a mortality rate of up to 40%. In this paper, we proposed an automatic segmentation network (CHSNet) to segment the lesions in cranial CT images based on the characteristics of acute cerebral hemorrhage images, such as high density, multi-scale, and variable location, and realized the three-dimensional (3D) visualization and localization of the cranial lesions after the segmentation was completed. To enhance the feature representation of high-density regions, and capture multi-scale and up-down information on the target location, we constructed a convolutional neural network with encoding-decoding backbone, Res-RCL module, Atrous Spatial Pyramid Pooling, and Attention Gate. We collected images of 203 patients with acute cerebral hemorrhage, constructed a dataset containing 5998 cranial CT slices, and conducted comparative and ablation experiments on the dataset to verify the effectiveness of our model. Our model achieved the best results on both test sets with different segmentation difficulties, test1: Dice = 0.918, IoU = 0.853, ASD = 0.476, RVE = 0.113; test2: Dice = 0.716, IoU = 0.604, ASD = 5.402, RVE = 1.079. Based on the segmentation results, we achieved 3D visualization and localization of hemorrhage in CT images of stroke patients. The study has important implications for clinical adjuvant diagnosis.
Classification of intracranial hemorrhage CT images based on texture analysis using ensemble-based machine learning algorithms: A comparative study
2023, Biomedical Signal Processing and Control
Blood vessels in the brain tissue can leak or burst, causing an intracranial hemorrhage, a potentially fatal disorder. One of the critical steps in brain stroke imaging is the categorization of intracranial hemorrhages. This article aims to perform texture-based intracranial hemorrhage computed tomography image classification with ensemble and machine learning classifiers using local binary pattern, local ternary pattern, and Weber local descriptor approach. The most common texture feature extraction method is a local binary pattern; however, other methods such as local ternary pattern, Weber local descriptor, and local binary pattern are also utilized for texture feature extraction. For all the extracted texture codes, histograms are applied, and finally, single feature vector texture codes by performing the concatenation of the bin. The efficacy of feature descriptors for classifying images into abnormal and normal classes was assessed using ten distinct classifiers. The result shows that Weber's local descriptor shows promising results in texture code classification for ensemble-based classification. For all texture-based methods, classification accuracy was better than standard machine learning. According to the findings of the experiment, the Random Forest classifier outperformed all other classifiers in terms of classification accuracy (86.55%), recall (86.31%), precision (87.23%), sensitivity (86.31%), specificity (86.81%), and F1-score (86.77%) for the Weber local descriptor.
Quality assessment of stroke radiomics studies: Promoting clinical application
2023, European Journal of Radiology
To evaluate the quality of radiomics studies on stroke using a radiomics quality score (RQS), Minimum Information for Medial AI reporting (MINIMAR) and Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD) to promote clinical application.
PubMed MEDLINE and Embase were searched to identify radiomics studies on stroke. Of 464 articles, 52 relevant original research articles were included. The RQS, MINIMAR and TRIPOD were scored to evaluate the quality of the studies by neuroradiologists.
Only four studies (7.7 %) performed external validation. The mean RQS was 3.2 of 36 (8.9 %), and the basic adherence rate was 24.9 %. The adherence rate was low for conducting phantom study (1.9 %), stating comparison to ‘gold standard’ (1.9 %), offering potential clinical utility (13.5 %) and performing cost-effectiveness analysis (1.9 %). None of the studies performed a test–retest, stated biologic correlation, conducted prospective studies, or opened codes and data to the public, resulting in low RQS. The total MINIMAR adherence rate was 47.4 %. The overall adherence rate for TRIPOD was 54.6 %, with low scores for reporting the title (2.0 %), key elements of the study setting (6.1 %), and explaining the sample size (2.0 %).
The overall radiomics reporting quality and reporting of published radiomics studies on stoke was suboptimal. More thorough validation and open data are needed to increase clinical applicability of radiomics studies.
Differentiation and prediction of pneumoconiosis stage by computed tomography texture analysis based on U-Net neural network
2022, Computer Methods and Programs in Biomedicine
Citation Excerpt :
Manual segmentation is a time-consuming method, which is inconvenient for rapid diagnosis. Therefore, texture analysis based on deep learning has become a novel research direction [14–17]. Traditional segmentation algorithms such as threshold segmentation, edge algorithm, and region growth can only be used to deal with simple scenes [18–20].
The progressive worsening of pneumoconiosis will ensue a hazardous physical condition in patients. This study details the differential diagnosis of the pneumoconiosis stage, by employing computed tomography (CT) texture analysis, based on U-Net neural network.
The pneumoconiosis location from 92 patients at various stages was extracted by U-Net neural network. Mazda software was employed to analyze the texture features. Three dimensionality reduction methods set the best texture parameters. We applied four methods of the B11 module to analyze the selected texture parameters and calculate the misclassified rate (MCR). Finally, the receiver operating characteristic curve (ROC) of the texture parameters was analyzed, and the texture parameters with diagnostic efficiency were evaluated by calculating the area under curve (AUC).
The original film was processed by Gaussian and Laplace filters for a better display of the segmented area of pneumoconiosis in all stages. The MCR value obtained by the NDA analysis method under the MI dimension reduction method was the lowest, at 10.87%. In the filtered texture feature parameters, the best AUC was 0.821.
CT texture analysis based on the U-Net neural network can be used to identify the staging of pneumoconiosis.

View all citing articles on Scopus

¹: Yu Liu and Qiong Fang are co-first authors.

View full text