Intelligent Algorithm-Based MRI Image Features for Evaluating the Effect of Nursing on Recovery of the Neurological Function of Patients with Acute Stroke

The aim of this study is to analyze the application of early rehabilitation nursing in nursing intervention of neurological impairment among patients with acute ischemic stroke. 116 patients with acute ischemic stroke were selected as the research subjects in this paper. The patients were divided into 58 experimental (early rehabilitation care) and 58 control (routine rehabilitation care) groups according to the difference of care protocols, all of which were performed magnetic resonance imaging on. An image resolution reconstruction algorithm on the basis of deep convolutional neural network is proposed for MRI image processing. The results show that peak signal to noise ratio (PSNR) and structural similarity index measure (SSIM) of the included algorithm were remarkably greater than those of compressed sensing (CS) algorithm and nonlocal similarity and block low rank prior-based NSBL algorithm. Running time was shorter than that of the latter two algorithms (P < 0.05). The neurological impairment scores of patients in the experimental group 3 and 5 weeks after treatment were obviously lower than those of patients in the control group (P < 0.05). The Barthel indexes of patients in the experimental group 3 and 5 weeks after treatment were obviously higher than those of patients in the control group (P < 0.05). FugI-Meyer assessment (FMA) and Disability of Arm-Shoulder-Hand (DASH) scores of patients in the experimental group 3 and 5 weeks after treatment were obviously lower than those of patients in control group (P < 0.05). The results show that the deep learning algorithm for MRI image processing performance is better than the traditional algorithm. It not only improves the image quality but also improves the processing efficiency. Early rehabilitation nursing and routine rehabilitation nursing can effectively improve the neurological deficit symptoms, limb motor function, and daily living ability of patients with acute ischemic stroke, and the effect of early rehabilitation nursing is the best.


Introduction
Acute stroke is due to various reasons that lead to blood supply disorders in local brain tissue, leading to ischemia and hypoxia in brain tissue and necrosis, resulting in corresponding symptoms and signs. It is a disease with high incidence, high recurrence rate, high death rate, high disability rate, and many complications. e common symptoms include weakness, numbness, or paralysis on one side or upper and lower limbs, weakness, numbness or paralysis on the face, blurred vision on one or both eyes, difficulty in language expression, dizziness, imbalance, and headache [1][2][3]. After treatment, about 70% of stroke patients have varying degrees of dysfunction. e severity of dysfunction depends on the damage to the brain structure [4,5]. Clinical treatment of stroke mainly includes intravenous thrombolysis, vascular interventional therapy, antiplatelet anticoagulation and expansion, and improvement of cerebral vascular circulation. Conservative treatment and surgical treatment are also considered for acute stroke [6]. Statistics show that China has become one of the countries with the highest incidence of stroke. Its annual economic losses have reached more than 40 billion yuan [7]. Each year, the number of new patients reaches more than 2 million, of which ischemic stroke accounts for 70%. About 70%-80% of ischemic stroke patients lead to disability and cannot live independently [8]. With the aging of population, the development of the economic level and the change of lifestyle, the intervention of early nursing and rehabilitation therapy for patients with ischemic stroke is particularly important, so it is necessary to discuss.
Neuroimaging plays an important role in acute stroke. For example, to determine or exclude the diagnosis of cerebrovascular disease (ischemic or hemorrhagic) and to provide evidence of possible stroke pathogenesis (embolism, hemodynamics, etc.). rough the size and location of the lesion and vascular state, we provide important information about the prognosis and so on [9][10][11]. CT examination can be used as an accurate screening method. e differential point is that clinicians collect medical history and preliminarily determine that patients are involved in stroke after physical examination. erefore, head CT is used in the first screening, which has the advantages of short time consumption and low cost [12]. MRI is superior to CT in differentiating acute ischemic stroke. e latest TTA report of the American Society of Neurology points out that DWI is determined to be useful for the diagnosis of acute ischemic stroke within 12 h of onset and should be more useful than nonenhanced CT [13,14]. In addition, multiparameter MRI may prolong the treatment time window. e basis is the understanding of individual pathophysiological processes. MRI is used to identify patients most likely to benefit from treatment (those with ischemic penumbra or risk tissue) and those most likely to have complications (such as hemorrhagic transformation, malignant infarction) [15]. In clinical practice, due to the limitation of acquisition time, patient comfort, image signal-to-noise ratio, and other factors, it is often difficult to obtain ideal high-resolution magnetic resonance images. It is easy to affect the doctor's objective judgment [16]. At present, with the rapid development of deep learning, deep networks for image super-resolution problems emerge endlessly and have achieved very good results. Super-resolution (SR) reconstruction technology can reconstruct images with high resolution from low-resolution images, which has broad application prospects in the field of magnetic resonance. However, the traditional SR method belongs to supervised learning, which requires specific training data for image reconstruction with poor reconstruction effects.
In summary, the nursing treatment of neurological deficit symptoms in stroke patients still needs further discussion. is study selects 116 patients with acute ischemic stroke as the research subject. According to different nursing schemes, the patients were divided into 58 cases of experimental group (early rehabilitation nursing) and 58 cases of control group (routine rehabilitation nursing). All patients underwent MRI examination. An image resolution reconstruction algorithm on the basis of deep learning is proposed for MRI image processing. e therapeutic effect of early nursing rehabilitation on patients was comprehensively evaluated by comparing the neurological deficit symptoms, limb motor function, and daily living ability scores of the two groups before and after treatment.

Research Subjects.
In this study, 116 patients with acute ischemic stroke admitted to hospital from June 1, 2019 to February 1, 2021 were selected as subjects. ere were 72 males and 44 females aged 48-71 years. is study has been approved by the ethics committee of hospital. e family members of the patients signed the informed consent.

Case Grouping and
Nursing. According to different nursing schemes, the patients were divided into 58 cases of the experimental group (early rehabilitation nursing) and 58 cases of the control group (routine rehabilitation nursing).
Two groups of patients were responsible for rehabilitation training. Training began after the condition was stable and neurological symptoms were stable. Specific steps were as follows: (1) Teaching patients to make limbs in functional position. Family members and nursing staff adopted the correct posture for counseling: affected side position-> healthy side position > supine position. Guide the patient to turn over every 2 hours and take back count. Avoid forming abnormal patterns of upper limb flexion, lower limb extension, and foot sag varus. (2) e hands and fingers crossed and held each other. e thumbs of the affected side were placed above, and the healthy limbs were used to drive the diseased limbs. en, slowly put down the arms and placed in front of the abdomen. Upper limb joint activities include finger nose, lateral lifting, lower limb flexion brace bed hip lifting, namely, bridge movement. (3) Patients were encouraged to sit for training as early as possible after their condition stabilized. According to the patient' s muscle strength, patients should take the initiative as much as possible. Methods of family protection: the patient was moved laterally to the bedside, and the affected leg was hooked with a healthy leg to move out of the bed, making the affected knee buckled to the bedside. en, the body rotates to the affected side, and the healthy hand crosses the body. At the same time, the healthy arm strength pushed the bed to sit up. When the torso had certain control ability, it can gradually let the patient use the healthy hand to assist the patient to picked up or dropped the object. en, gradually put out to increase the difficulty of sitting training. (4) Lower limb muscle strength grade 3 and above for standing training. When standing training, the therapist's protection method was as follows: the upper limbs forward, head and trunk forward, the center of gravity moved between the feet. en, raise the patient's hips and knees were gradually stretched and made to stand up. e patient held the parallel bars by hand, gradually increased the time and tried to shift the center of gravity to the affected side. Moreover, gradually increased the standing balance training. (5) With the increase of lower limb weight-bearing muscle strength, walking training can be gradually carried out. en, gradual transition to across different obstacles and stairs walking training was given. It mainly increases brain blood supply, activates brain cells, and improves local muscle spasm, twitch, and incomplete paralysis so as to improve muscle strength and motor function. e above rehabilitation exercise treatment was carried out every 20-30 min, 2-3 times a day, 4-5 days a week. Rehabilitation exercise therapy lasted 3 months.
Both groups were given drug treatment at the same time. Drug therapy was conducted in accordance with acute ischemic cerebrovascular disease diagnosis and treatment guidelines. Basic medications include aspirin, ginkgo dipyridamole, and edaravone.

MRI Examination.
e patients were examined by 3.0 T superconducting magnetic resonance scanning, and the 12channel phased array head coil was used. Gradient echo sequence of magnetization preparation, scanning parameters were as follows: TE 900 ms, TR 2,200 ms, FOV 256 × 256 × 150, 1 mm voxel collection. Planar echo imaging sequence, scanning parameters were as follows: TE 40 ms, TR 3,000 ms, FOV 215 × 215, 2 mm voxel acquisition, inversion angle 90°, scanning time 320 s. Finally, the obtained image data was input to the workstation for enhancement processing.

Deep Convolutional Neural Network (CNN)-Based Image
Resolution Reconstruction Algorithm. With the rapid development of deep learning, deep networks for image superresolution problems emerge in endlessly and have achieved very good results. However, most super-resolution methods on the basis of deep learning belong to supervised learning, which limits the reconstruction process to specific training data. e basic framework of these methods is illustrated in Figure 1. e specific downsampling method was used to process the high-resolution images to obtain the corresponding low-resolution image, and then the corresponding training data were collected. A deep network model was trained.
en, the low-resolution image to be tested was input into the model to obtain the reconstructed superresolution image.
e purpose of super-resolution reconstruction is to recover high-resolution images from low-resolution images. Set the low-resolution image to P, which can be expressed as follows: where C represents the degradation process function, P y represents the high-resolution image, θ represents the degradation process parameters. en, the downsampling operation can be expressed as follows: ↓ r represents the downsampling operation of r, and r { } ∈ θ. In general, low-resolution images are obtained by bicubic anti-aliasing reduction of high-resolution images, so the degradation process can also be expressed as follows: where β represents fuzzy kernel, P y ⊗ β represents convolution operation between high-resolution image and fuzzy kernel, and u τ represents Gaussian white noise. e objective function of super-resolution reconstruction can be expressed as follows: where loss(P y * , P y ) represents the loss function and Φ(α) represents the regularization term. e above supervised image super-resolution algorithm can obtain good results. But a lot of paired training data are needed to train the network. For magnetic resonance imaging, which is limited by hardware conditions, it is difficult to obtain such data sets. erefore, this paper regards super-resolution image reconstruction as an ill-posed inverse problem, that is, low- resolution images correspond to multiple different highresolution images. Considering the actual operation process, image quality will be affected by many aspects. For example, the motion artifacts caused by the movement of the subjects in the data acquisition process, the black band artifacts caused by the inhomogeneity of the magnetic field, and the noise caused by the current fluctuation in the circuit, etc. So, introducing these factors into image degradation model can be expressed as follows: where P L represents a low-resolution image, T represents degradation function, C down represents the downsampling operator, and χ represents noise.
In magnetic resonance imaging, the degradation function can be seen as a part of the k-space center of the original high-resolution magnetic resonance image that intercepts the corresponding proportion. erefore, a low-resolution magnetic resonance image is given to recover the image version as similar as the real collected high-resolution image.
is study reconstructs super-resolution images by minimizing the following loss functions: where P * H represents the reconstruction of high-resolution images. e above formulation often leads to an unstable solution. So, it is necessary to give a suitable regularization term to limit the range of solution space, then the model can be updated to as follows: TIME represents the spin locking time, l 0 represents the magnetization in equilibrium state, and P σ represents the parameters to be fitted. e weighted images collected at different spin locking times were fitted by the second exponential decay model. arg min P H ‖P L − C down T H ‖ 2 ensures data consistency. ‖P H − l 0 e − (TIME/P ω ) ‖ 2 ensures model consistency. Since the purpose of network learning is to learn the mapping from low-resolution images to highresolution images, the above can also be expressed as follows: where (C down T) − 1 represents the combination of the recovery operator and the sampling on the image.

Image Quality Assessment Indicators.
In this study, peak signal-to-noise ratio (PSNR) [17] and structural similarity (SSIM) [18] are used as two parameters to evaluate image quality. PSNR can be defined by the maximum pixel value and mean square error between images, which can be expressed as follows: where M represents pixels, P * represents the reconstructed image, and P represents the original image. SSIM is used to measure the structural similarity between images.
e evaluation criteria are based on the brightness, contrast, and structure of the image, which can be expressed as follows: SSIM � R 1 P, P * a R 2 P, P * b R 3 P, P * c , where υ P represents image brightness and ς P represents contrast ratio.

Observation
Indicators. e general data (age, gender, course of disease) of the two groups were recorded. e functional indexes before treatment, 3 weeks after treatment, and 5 weeks after treatment were recorded. e method suggested by National Institutes of Health Stroke (NIHSS) was used to assess the degree of neurological deficits. Activities of daily living were assessed by the modified Barthel Index (MBI). FugI-Meyer Assessment (FMA) was used to assess limb motor function. e DASH (disability of Arm-Shoulder-Hand) scale was used to score the patients before and after treatment [19].

Statistical Methods.
e data in this study were analyzed by SPSS19.0. e measurement data are expressed as mean ± standard deviation ( ± s). Count data are expressed as percentage (%). One-way ANOVA was used for pairwise comparison. e difference was statistically significant with P < 0.05.

Algorithm Performance Display.
e MRI image reconstruction algorithm based on compressed sensing (CS) [20], the MRI image reconstruction algorithm based on nonlocal similarity and block low rank prior (NSBL) [21] are introduced, and compared with the algorithms designed in this study. It can be seen from the actual MRI image reconstruction ( Figure 2) that the overall quality of the original image is poor, there are more artifacts and noises, and the resolution is low, and the clarity is obviously not enough. After the three algorithms, the quality has been significantly improved. Among them, the algorithm in this study has the best re-effect on MRI images, the clarity has been significantly improved, and the artifacts and noise have been greatly reduced.
From the quantitative data results (Figure 3), the PSNR and SSIM of the proposed algorithm are significantly greater than those of the CS algorithm and the NSBL algorithm. e difference was statistically significant (P < 0.05). e running time of the proposed algorithm is significantly shorter than that of the CS algorithm and NSBL algorithm. e difference was statistically significant (P < 0.05).

Comparison of Basic Data between the Two Groups of Patients.
e basic data of the two groups were compared (Figure 4). ere was no significant difference in age, gender, course of disease, location of disease (left hemisphere, right hemisphere, brainstem), and degree of nerve deficit (mild, moderate, and severe) between the two groups (P > 0.05). Figure 5, the patient was a 59-year-old woman. e patient was treated for weakness and rigidity of the right lower limb. ADC map showed the low-signal region at the right parietaloccipital junction, FLAIR image showed a high signal in the corresponding region, T1WI showed a low signal in the corresponding region, and T1-enhanced scan showed brain parenchyma enhancement in the affected region. Figure 6 reveals that the patient is a 70-year-old male. DWI displayed the low-signal region of the right occipital lobe, and the high signal edge was visible around it. is may be caused by the T2 penetration effect. e ADC map revealed a high signal in the corresponding region, SWI revealed the results of right occipital lobe hemorrhage, and T2WI revealed a high signal in the corresponding region of right occipital lobe. Figure 7 shows that the difference in neurological deficit scores between the two groups before treatment was not statistically significant (P > 0.05). e neurological deficit scores of the two groups at 3 and 5 weeks after treatment were significantly lower than those before treatment, and the difference was statistically significant (P < 0.05). e neurological deficit score of the experimental  Contrast Media & Molecular Imaging group was significantly lower than that of the control group at 3 and 5 weeks after treatment, and the difference was statistically significant (P < 0.05).

Comparison of the Barthel Index before and after
Treatment between the Two Groups. Figure 8 shows that the difference in the Barthel index between the two groups before treatment was not statistically significant (P > 0.05). e Barthel index of the two groups at 3 and 5 weeks after treatment was significantly higher than that before treatment, and the difference was statistically significant (P < 0.05). e Barthel index of the experimental group was significantly higher than that of the control group at 3 and 5 weeks after treatment, and the difference was statistically significant (P < 0.05).

Comparison of Motor Function Scores between the Two
Groups before and after Treatment. Figure 9 shows that the  FMA difference between the two groups before treatment was not statistically significant (P > 0.05). FMA of the two groups at 3 and 5 weeks after treatment was significantly lower than that before treatment, and the difference was statistically significant (P < 0.05). FMA of the experimental group at 3 and 5 weeks after treatment was significantly lower than that of the control group, and the difference was statistically significant (P < 0.05).

Comparison of DASH between the Two Groups before and
after Treatment. Figure 10 shows that the difference in DASH between the two groups before treatment was not statistically significant (P > 0.05). e DASH of the two groups at 3 and 5 weeks after treatment was significantly lower than that before treatment, and the difference was statistically significant (P < 0.05). e DASH of the experimental group was significantly lower than that of the control group at 3 and 5 weeks after treatment, and the difference was statistically significant (P < 0.05). Experimental group Control group Figure 7: Two groups of patients before and after treatment neurological deficit score 0 is before treatment; 1 is three weeks after treatment; and 2 is five weeks after treatment. # Compared with that before treatment, P < 0.05; #compared with the experimental group, P < 0.05.   Experimental group Control group Figure 10: DASH scores of two groups before and after treatment. 0 is before treatment. 1 is three weeks after treatment. 2 is 5 weeks after treatment. * Compared with that before treatment, P < 0.05; # compared with the experimental group, P < 0.05.

Discussion
Stroke is caused by a sudden rupture of blood vessels in the brain bleeding or cerebral vascular embolism caused by blood that cannot flow into the brain due to brain tissue damage. Stroke can lead to cerebral ischemia, hypoxia, necrosis, and other pathological changes, which lead to neurological deficits and other neurological symptoms. Its severity will directly affect the prognosis of patients, which is one of the topics of widespread concern in clinical practice [22]. erefore, 116 patients with acute ischemic stroke were selected as the research subjects in this study. According to different nursing schemes, the patients were divided into 58 cases of the experimental group (early rehabilitation nursing) and 58 cases of the control group (routine rehabilitation nursing). e patients were examined by Avanto3.0 T superconducting magnetic resonance scanning in Siemens, Germany. e PSNR and SSIM of the proposed algorithm are significantly larger than those of the CS algorithm and NSBL algorithm. e running time was significantly shorter than that of the CS algorithm and NSBL algorithm, and the difference was statistically significant (P < 0.05).
is is similar to the research results of Kopczak et al. (2020) [23]. It shows that the deep learning algorithm constructed in this paper has better processing performance than the traditional algorithm for MRI images, which not only improves the image quality but also improves the processing efficiency. e basic data of the two groups were compared, and it can be seen that there was no statistically significant difference in age, gender, course of disease, location of disease (left hemisphere, right hemisphere, and brainstem), and degree of nerve deficit (mild, moderate, and severe) between the two groups (P > 0.05). is provides feasibility for subsequent research.
e neurological deficit scores of the two groups at 3 and 5 weeks after treatment were significantly lower than those before treatment. e neurological deficit scores of the experimental group at 3 and 5 weeks after treatment were significantly lower than those of the control group, and the difference was statistically significant (P < 0.05).
is is consistent with the research results of Lu et al. (2020) [24]. It indicated that early rehabilitation nursing and routine rehabilitation nursing could effectively improve the neurological deficit symptoms of patients with acute ischemic stroke, and the effect of early rehabilitation nursing was better. e Barthel index of the two groups at 3 and 5 weeks after treatment was significantly higher than that before treatment. e Barthel index of the experimental group was significantly higher than that of the control group at 3 and 5 weeks after treatment, and the difference was statistically significant (P < 0.05). is also reveals that early rehabilitation nursing can more effectively improve the daily living ability of stroke patients [25]. e FMA and DASH scores of the two groups at 3 and 5 weeks after treatment were significantly lower than those before treatment. e FMA and DASH scores of the experimental group were significantly lower than those of the control group at 3 and 5 weeks after treatment, and the difference was statistically significant (P < 0.05). is reveals that early rehabilitation nursing can more effectively improve the limb motor ability of stroke patients.

Conclusion
In this study, 116 patients with acute ischemic stroke were selected as subjects. According to different nursing schemes, the patients were divided into 58 cases of the experimental group (early rehabilitation nursing) and 58 cases of the control group (routine rehabilitation nursing). Patients were examined by magnetic resonance imaging on the basis of deep learning algorithm.
e results show that the deep learning algorithm constructed in this paper has better processing performance than the traditional algorithm for MRI images, which not only improves the image quality but also improves the processing efficiency. Early rehabilitation nursing and routine rehabilitation nursing can effectively improve the neurological deficit symptoms, limb motor function, and daily living ability of patients with acute ischemic stroke, and the effect of early rehabilitation nursing is the best. However, only 116 patients with acute cerebral stroke were included. e sample size was small with single source. Besides, there was apparent regional limitations. erefore, a wider range of cases needed to be included in subsequent studies to further discuss clinical nursing intervention paths in ischemic stroke. In summary, this study provides a theoretical support for the clinical nursing treatment of patients with ischemic stroke.

Data Availability
e data used to support the findings of this study are available from the corresponding author upon request.

Conflicts of Interest
e authors declare that they have no conflicts of interest.