Application of Hyperspectral Imaging in the Diagnosis of Glomerular Diseases

Morning urine samples of patients with IgA nephropathy, idiopathic membranous nephropathy, diabetic nephropathy, and minimal degenerative nephropathy confirmed by renal needle biopsy in our hospital from November 2020 to January 2022 were collected. By scanning the morning urine samples of corresponding patients with microhyperspectral imager, machine learning and deep learning were used to classify microhyperspectral images, and the classification accuracy was greater than 85%. Thus, hyperspectral imaging technology could be used as a non-invasive diagnostic means to assist the diagnosis of glomerular diseases.

Study Overview

• Status: Not yet recruiting
• Conditions: Diabetic Nephropathies; Glomerulonephritis, IGA; Nephrosis, Lipoid; Glomerulonephritis, Membranous; Hyperspectral Imaging
• Intervention / Treatment: Diagnostic test: Microscopic hyperspectral imaging system

• Study Type: Observational
• Enrollment (Anticipated): 80

Contacts and Locations

• Study Contact: Name: Wang Zongsong; Phone Number: 18660190175; Email: wzsong3@163.com

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: 18 years to 90 years (Adult, Older Adult)
• Accepts Healthy Volunteers: No

• Sampling Method: Probability Sample

Study Population

Patients with massive proteinuria were diagnosed as IgA nephropathy, idiopathic membranous nephropathy, diabetic nephropathy and minimal change nephropathy by renal biopsy.

Description

Inclusion Criteria:

• Over 18 years old;
• Patients with IgA nephropathy, idiopathic membranous nephropathy, diabetic nephropathy, minimal change nephropathy confirmed by renal biopsy;
• Had not received hormone and/or immunosuppressive therapy before renal biopsy;
• Complete clinical data, all signed the "Admission Certificate of Qianfoshan Hospital of Shandong Province", and agreed to use relevant medical information, biological specimen examination and examination results for scientific research.

Exclusion Criteria:

• There are factors causing secondary membranous nephropathy, such as immune diseases (systemic lupus erythematosus), tumors/infections (viral hepatitis), drugs or poisons, etc.;
• Severe infection: fever, cough and expectoration, sore throat, abdominal pain, diarrhea, carbuncle and furuncle and other clinical manifestations of skin and soft tissue infection, blood routine white blood cell count beyond the normal range (10×109/L);
• Severe cardiovascular disease: including chronic heart failure grade 3 or above and various arrhythmias;
• Infectious diseases: active hepatitis, AIDS, syphilis, etc. ;
• Tumor evidence: it has been found that there is a certain tumor or clinical manifestations, tumor markers, etc., suggesting the possibility of tumor.

Study Plan

How is the study designed?

Number of groups / cohorts

4

Cohorts and Interventions

Group / Cohort	Intervention / Treatment
diabetic nephropathy; Urine samples were collected from patients with IgA nephropathy, idiopathic membranous nephropathy, diabetic nephropathy and minimal change nephropathy. The samples were centrifuged and frozen in a refrigerator at -80 degrees Celsius. The images were divided into a training set and a test set at a fixed ratio. The digital images were input into classification models such as one-dimensional convolutional neural network to learn and test. The training set was used for the training and parameter iteration of the artificial intelligence non-invasive fluid diagnosis model, and the test set was used for the recognition and interpretation of the model. The confusion matrix, accuracy and ROC curve were calculated through the interpretation results to evaluate the performance of the model.	Diagnostic test: Microscopic hyperspectral imaging system; Microscopic hyperspectral imaging system
minimal change nephropathy; Urine samples were collected from patients with IgA nephropathy, idiopathic membranous nephropathy, diabetic nephropathy and minimal change nephropathy. The samples were centrifuged and frozen in a refrigerator at -80 degrees Celsius. The images were divided into a training set and a test set at a fixed ratio. The digital images were input into classification models such as one-dimensional convolutional neural network to learn and test. The training set was used for the training and parameter iteration of the artificial intelligence non-invasive fluid diagnosis model, and the test set was used for the recognition and interpretation of the model. The confusion matrix, accuracy and ROC curve were calculated through the interpretation results to evaluate the performance of the model.	Diagnostic test: Microscopic hyperspectral imaging system; Microscopic hyperspectral imaging system
IgA nephropathy; Urine samples were collected from patients with IgA nephropathy, idiopathic membranous nephropathy, diabetic nephropathy and minimal change nephropathy. The samples were centrifuged and frozen in a refrigerator at -80 degrees Celsius. The images were divided into a training set and a test set at a fixed ratio. The digital images were input into classification models such as one-dimensional convolutional neural network to learn and test. The training set was used for the training and parameter iteration of the artificial intelligence non-invasive fluid diagnosis model, and the test set was used for the recognition and interpretation of the model. The confusion matrix, accuracy and ROC curve were calculated through the interpretation results to evaluate the performance of the model.	Diagnostic test: Microscopic hyperspectral imaging system; Microscopic hyperspectral imaging system
idiopathic membranous nephropathy; Urine samples were collected from patients with IgA nephropathy, idiopathic membranous nephropathy, diabetic nephropathy and minimal change nephropathy. The samples were centrifuged and frozen in a refrigerator at -80 degrees Celsius. The images were divided into a training set and a test set at a fixed ratio. The digital images were input into classification models such as one-dimensional convolutional neural network to learn and test. The training set was used for the training and parameter iteration of the artificial intelligence non-invasive fluid diagnosis model, and the test set was used for the recognition and interpretation of the model. The confusion matrix, accuracy and ROC curve were calculated through the interpretation results to evaluate the performance of the model.	Diagnostic test: Microscopic hyperspectral imaging system; Microscopic hyperspectral imaging system

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Microhyperspectral image of urine specimen	Microhyperspectral images of urine samples from patients with four different glomerular diseases before treatment	2023.4-2023.10

Collaborators and Investigators

• Sponsor: Qianfoshan Hospital

Study record dates

Study Major Dates

• Study Start (Anticipated): May 30, 2023
• Primary Completion (Anticipated): August 20, 2023
• Study Completion (Anticipated): September 20, 2023

Study Registration Dates

• First Submitted: March 18, 2023
• First Submitted That Met QC Criteria: March 31, 2023
• First Posted (Actual): April 4, 2023

Study Record Updates

• Last Update Posted (Actual): May 3, 2023
• Last Update Submitted That Met QC Criteria: April 29, 2023
• Last Verified: April 1, 2023

More Information

Terms related to this study

• Additional Relevant MeSH Terms: Immune System Diseases; Autoimmune Diseases; Kidney Diseases; Urologic Diseases; Endocrine System Diseases; Diabetes Complications; Diabetes Mellitus; Nephritis; Diabetic Nephropathies; Glomerulonephritis; Glomerulonephritis, IGA; Nephrosis; Glomerulonephritis, Membranous; Nephrosis, Lipoid
• Other Study ID Numbers: liquid biopsy-glomerulopathy

Drug and device information, study documents

• Studies a U.S. FDA-regulated drug product: No
• Studies a U.S. FDA-regulated device product: No