ObjectiveTo apply the multi-modal deep learning model to automatically classify the ultra-widefield fluorescein angiography (UWFA) images of diabetic retinopathy (DR). MethodsA retrospective study. From 2015 to 2020, 798 images of 297 DR patients with 399 eyes who were admitted to Eye Center of Renmin Hospital of Wuhan University and were examined by UWFA were used as the training set and test set of the model. Among them, 119, 171, and 109 eyes had no retinopathy, non-proliferative DR (NPDR), and proliferative DR (PDR), respectively. Localization and assessment of fluorescein leakage and non-perfusion regions in early and late orthotopic images of UWFA in DR-affected eyes by jointly optimizing CycleGAN and a convolutional neural network (CNN) classifier, an image-level supervised deep learning model. The abnormal images with lesions were converted into normal images with lesions removed using the improved CycleGAN, and the difference images containing the lesion areas were obtained; the difference images were classified by the CNN classifier to obtain the prediction results. A five-fold cross-test was used to evaluate the classification accuracy of the model. Quantitative analysis of the marker area displayed by the differential images was performed to observe the correlation between the ischemia index and leakage index and the severity of DR. ResultsThe generated fake normal image basically removed all the lesion areas while retaining the normal vascular structure; the difference images intuitively revealed the distribution of biomarkers; the heat icon showed the leakage area, and the location was basically the same as the lesion area in the original image. The results of the five-fold cross-check showed that the average classification accuracy of the model was 0.983. Further quantitative analysis of the marker area showed that the ischemia index and leakage index were significantly positively correlated with the severity of DR (β=6.088, 10.850; P＜0.001). ConclusionThe constructed multimodal joint optimization model can accurately classify NPDR and PDR and precisely locate potential biomarkers.
ObjectiveTo assess the reliability, effectiveness, and the safety of full endoscopic transforaminal decompression (FETD) under local anesthesia guided by the classification of lateral region of the lumbar spinal canal (CLLSC) in treating lumbar spinal stenosis (LSS) in geriatric patients.MethodsThe clinical data of 63 geriatric patients with LSS met the inclusion criteria underwent FETD surgery between June 2015 and July 2017 were retrospectively analyzed. There were 37 males and 26 females, with a median age of 76 years [interquartile range (IQR), 73-80 years], and a median symptomatic duration of 55 months (IQR, 16-120 months). There were 17 cases of grade B and 46 cases of grade C based on the Schizas morphological grading system. CLLSC was used for imaging evaluation for the stenotic condition, and intra-class correlation coefficients (ICC) were used to test intra-observer and inter-observer reliability of CLLSC. The stenotic condition of patients was re-evaluated by the surgeon after operation, and the results were compared with the findings of preoperative CLLSC. The visual analogue scale (VAS) score for low back pain and leg pain recorded before operation, and at 1 day, 3 months, and 6 months after operation, and last follow-up were used to assess the pain relieving; the functional improvement was evaluate by Oswestry disability index (ODI); the modified Macnab criteria were used to self-evaluate the surgical satisfaction.ResultsThe operation were successfully performed for all patients, with a median operation time of 75 minutes (IQR, 65-85 minutes), postoperative hospitalization stay of 48 hours (IQR, 48-72 hours), and the time to ambulation after operation of 24 hours (IQR, 24-24 hours). Sixty-three patients were followed-up and with a median follow-up time of 18 months (IQR, 13-20 months). Based on preoperative CLLSC classification, there were 72 stenotic zones, distributed 16 in zone 1, 6 in zone 2, 3 in zone 3, 2 in zone 4, 7 in zone 5, 34 in zones 1+2, 2 in zones 3+4, and 2 in zones 4+5. Perioperative complications occurred in 4 cases (6.3%), including 2 cases of intraoperative dural sac tear, 1 of preoperative numbness symptom aggravation, and 1 of postoperative urinary retention. VAS score of leg pain and ODI score at each time point after operation were significantly improved compared with those before operation (P<0.05). VAS scores of low back pain showed no significant difference between pre- and post-operation (P>0.05). At last follow-up, based on the modified Macnab criteria, 19 cases were excellent, 37 were good, 6 were fair, and 1 was poor, and the excellent and good rate was 88.9%. The reliability analysis showed that CLLSC had substantial intra-observer reliability in the geriatric population, with an average ICC of 0.78. There was also a substantial inter-observer reliability, with an average ICC of 0.73. While comparing the preoperative CLLSC results with the postoperative CLLSC results, 53 patients (73.6%) were in full agreement, 15 patients (20.8%) were in partial agreement, and 4 patients (5.6%) were not.ConclusionCLLSC has high reliability in the diagnosis of LSS in the geriatric patients. Combined FETD with CLLSC, accurate diagnosis, and minimal invasion can be performed to achieve safe and effective result.
ObjectiveTo investigate the value of gadolinium ethoxybenzyl diethylenetriamine pentaacetic acid enhanced magnetic resonance imaging (Gd-EOB-DTPA-MRI) in gross morphological classification of hepatocellular carcinoma (HCC). MethodsThe clinicopathologic data of patients with HCC who received surgical treatment in the Affiliated Huai’an Hospital of Xuzhou Medical University from January 2017 to December 2022 were retrospectively gathered. The Gd-EOB-DTPA-MRI was performed before operation. Two radiologists independently assessed the gross morphological classification of HCC according to the imaging performance. The tumors were cut into sections in a coronal plane and were taken pictures for recording pathological features after operation. The tumors were assigned into 4 types according to the references and clinical experiences: single nodular type (SN), single nodular with extranodular growth type (SN-EG), confluent multi-nodular type (CMN), and infiltration type (IF). Matching degree of morphological classification was analyzed between by the Gd-EOB-DTPA-MRI and resected specimen. The pathological features of 4 types of HCC were also analyzed. ResultsA total of 87 patients with HCC were included. The gross morphological classification by the Gd-EOB-DTPA-MRI was 28 (32.2%) patients with SN, 28 (32.2%) patients with SN-EG, 21 (24.1%) patients with CMN, 10 (11.5%) patients with IF, which by the resected specimen was 33 (37.9%) patients with SN, 24 (27.6%) patients with SN-EG, 21 (24.1%) patients with CMN, and 9 (10.4%) patients with IF in the 87 patients with HCC. The Kappa’s coefficient of agreement between the results of Gd-EOB-DTPA-MRI and postoperative resection specimens was 0.776 (P=0.199). There were statistical differences in the tumor diameter and microvascular invasion (MVI) among the 4 types of gross morphology classification (F=2.937, P=0.038; χ2=16.852, P=0.001), the MVI rate was highest and tumor diameter was biggest in the patients with IF among the 4 types of gross morphology classification (P<0.05). ConclusionsFrom the results of this study, the gross morphological classification of HCC is closely related to the tumor diameter and MVI. Results of Gd-EOB-DTPA-MRI and postoperative resection specimens in assessing the gross morphological classification are good agreement. Therefore, an accurate preoperative planning and better therapy strategy for the patients with HCC can be provided according to gross morphological classification by preoperative Gd-EOB-DTPA-MRI.
Objective To evaluate whether the classification of parathyroid can be used to evaluate how difficult it is that the parathyroid glands get preserved in situ during thyroid surgery. Methods Clinical date were retrospectively collected from the patients with thyroid nodules, who had undergone the initial thyroidectomy in the Department of Thyroid Surgery, West China Hospital of Sichuan University between January 2014 and June 2016. The number of parathyroid glands was counted according to the classification of parathyroid. It got comparative analysis that the rates of parathyroid glands in situ among the different types. Results A total of 996 patients were included in the study, and 3 269 pieces of parathyroid glands were identified. The mean number of parathyroid identification was 3.3 pieces. These parathyroid glands consisted of 77.5% (2 532/3 269) type A and 22.5% (737/3 269) type B. The rate of parathyroid glands in situ was 77.1% (1 951/2 532) in type A, and 80.7% (595/737) in type B, the difference was significant (P=0.03). And the rate of parathyroid glands in situ in type A1 was significantly higher than that in type A2 (80.5%vs 21.4%,P<0.001). The parathyroid of type A3 couldn’t get preserved in situ. The rate of superior parathyroid glands in situ in type B1 was higher than that in type A1 (97.5%vs 93.7,P<0.01). But the rate of inferior parathyroid glands in type B1 was closed to that in type A1 (62.2%vs 65.7%,P=0.23), and both the rates were significant less than that in type B2 (86.0%) and in type B3 (90.2%),PA1vs B2=0.001,PA1vs B3<0.001,PB1vs B2=0.004,PB1vs B3=0.001. Conclusion The classification of parathyroid can be used to evaluate effectively how difficult it is that the parathyroid glands get preserved in situ during thyroid surgery.
Objective To analyze the current status of day surgery in West China Hospital of Sichuan University and compare it with the Recommended Catalogue of Day Surgery (2022 Edition), in order to provide reference for the update and expansion of day surgery procedures in hospitals. Methods The day surgical procedures carried out in West China Hospital of Sichuan University from October 2009 to July 2023 were collected. The day surgical procedures were classified based on surgical specialties, types of procedures, methods of operation, and surgical grading, etc. The day surgical procedures in West China Hospital of Sichuan University were compared with those in the Recommended Catalogue of Day Surgery (2022 Edition). Results A total of 576 types of daytime surgeries were carried out in West China Hospital of Sichuan University, involving 15 specialties. A total of 662 types of day surgeries were recommend in the nationally recommended catalogue for daytime surgery, involving 11 specialties. Among the nationally recommended surgical procedures, West China Hospital of Sichuan University had carried out 233 types of surgical procedures, and 26 additional types of surgical procedures were carried out as outpatient surgeries. According to the classification of surgical difficulty, the Level Ⅱ and Ⅲ surgeries were mainly carried out in West China Hospital of Sichuan University, and the Level Ⅱ surgeries were mainly recommend in the nationally recommended catalogue for day surgery. Conclusions The high-quality development of day surgery in China should not be limited to the supplement of surgical procedures. It is necessary to strive to increase the proportion of minimally invasive and fourth-level surgeries while ensuring medical quality and safety, and improve the overall medical level of day surgery.
Identification of molecular subtypes of malignant tumors plays a vital role in individualized diagnosis, personalized treatment, and prognosis prediction of cancer patients. The continuous improvement of comprehensive tumor genomics database and the ongoing breakthroughs in deep learning technology have driven further advancements in computer-aided tumor classification. Although the existing classification methods based on gene expression omnibus database take the complexity of cancer molecular classification into account, they ignore the internal correlation and synergism of genes. To solve this problem, we propose a multi-layer graph convolutional network model for breast cancer subtype classification combined with hierarchical attention network. This model constructs the graph embedding datasets of patients’ genes, and develops a new end-to-end multi-classification model, which can effectively recognize molecular subtypes of breast cancer. A large number of test data prove the good performance of this new model in the classification of breast cancer subtypes. Compared to the original graph convolutional neural networks and two mainstream graph neural network classification algorithms, the new model has remarkable advantages. The accuracy, weight-F1-score, weight-recall, and weight-precision of our model in seven-category classification has reached 0.851 7, 0.823 5, 0.851 7 and 0.793 6 respectively. In the four-category classification, the results are 0.928 5, 0.894 9, 0.928 5 and 0.865 0 respectively. In addition, compared with the latest breast cancer subtype classification algorithms, the method proposed in this paper also achieved the highest classification accuracy. In summary, the model proposed in this paper may serve as an auxiliary diagnostic technology, providing a reliable option for precise classification of breast cancer subtypes in the future and laying the theoretical foundation for computer-aided tumor classification.
Objective To investigate the recovery status of people wounded in the Wenchuan earthquake. Method Data were retrospectively collected from administrative documents in the Bureau of Medical Affairs, Sichuan Provincial Health Department. The severity of injury was assessed by Injury Severity Score (ISS). The data were recorded by EXCEL software and descriptive analysis was conducted. Results Our analysis results of rehabilitation treatment through Feb. 5, 2009 shows that 27,080 of the 28,008 patients had been treated and discharged, for a discharge rate of 97.8%. There were 928 patients still in hospitals at that time, including 55 cases of traumatic brain injury, 163 cases of paraplegia, 260 amputees, and 449 cases of severe spine, pelvis and other fractures. Some amputees needed to receive replacement of artificial limbs or stump dressing operation and rehabilitation; most patients who were installed internal fixation needed to removal and post-rehabilitation. Conclusions The effectiveness of rehabilitation is significant. Our work in the next stage should focus on (1) continuing to improve the establishment of province’s rehabilitation capabilities and increasing capital investment; (2) enhancing training for medical rehabilitation practitioners in order to improve operational standards and service capabilities; (3) developing the wounded rehabilitation standards in later stages, conducting follow-up and functional training in order to maximize recovery and return to society; (4) increasing employment opportunities for disabled persons.
ObjectiveTo investigate the relationship between preoperative C-reactive protein (CRP)/albumin ratio (CAR) and National Institutes of Health (NIH) risk classification in patients with gastric stromal tumors.MethodsClinical data of 108 patients with gastric stromal tumors admitted to the First Affiliated Hospital of Kunming Medical University from February 2010 to November 2016 were retrospectively collected. With the median of CAR as the critical value, patients were divided into high CAR group (CAR>0.048) and low CAR group (CAR≤0.048). Then observed the general clinicopathological characteristics and survival status of patients with higher and lower CAR value.ResultsThere were significant differences in NIH classification, tumor diameter, and mitosis between the high CAR group and low CAR group (P<0.05). Compared with the low CAR group, the tumors in the high CAR group had larger diameter, higher mitotic figure, and higher NIH grade. Survival analysis showed that the prognosis of the low CAR group was better than that of the high CAR group (χ2=15.152, P<0.001).ConclusionsCAR is closely related to the malignant index and NIH risk classification of gastric stromal tumors. It can be used as an index for evaluating the malignant degree of gastric stromal tumors, and it is expected to be an important reference factor for clinical NIH risk classification and prognosis.
ObjectiveTo investigate the effect of indocyanogen green retention rate (ICGR) measurement and Child-Pugh classification of liver function in the preoperative evaluation of hepatic functional reservation. MethodsThe level of ICGR at 15 min (ICGR15) and the percentage of liver fibrosis in 103 patients with liver cancer were measured before hepatectomy, and the Child-Pugh classification of liver function was evaluated before and after the operations, and their connections were analyzed. ResultsAs the rise of Child-Pugh classification of liver function, the percentage of liver fibrosis increased gradually, there were significant differences between any two Child-Pugh classification (Plt;0.05). There was a linear correlation between the ICGR15 and the liver fibrosis percentage (rs=0.960, Plt;0.05). The value of preoperative ICGR15 in patients with postoperative Child-Pugh classification from grade A to grade B or grade B to grade C was gnificantly higher than that in patients with stabilization of Child-Pugh classification before and after operations (Plt;0.05). ConclusionICGR15 combined with Child-Pugh classification can improve the accuracy for the evaluation of preoperative hepatic functional reservation.
The classification of lung tumor with the help of computer-aided diagnosis system is very important for the early diagnosis and treatment of malignant lung tumors. At present, the main research direction of lung tumor classification is the model fusion technology based on deep learning, which classifies the multiple fusion data of lung tumor with the help of radiomics. This paper summarizes the commonly used research algorithms for lung tumor classification, introduces concepts and technologies of machine learning, radiomics, deep learning and multiple data fusion, points out the existing problems and difficulties in the field of lung tumor classification, and looks forward to the development prospect and future research direction of lung tumor classification.