Bladder Cancer Staging and Prediction of New Adjuvant Chemotherapy Efficacy Based on Deep Learning and Transfer Learning in Ultrasound-Magnetic Resonance-Pathology Multimodal Multiscale

Intelligent Diagnosis of Bladder Cancer Staging and Prediction of New Adjuvant Chemotherapy Efficacy Based on Deep Learning and Transfer Learning in Ultrasound-Magnetic Resonance-Pathology Multimodal Multiscale

Bladder cancer is the most common malignant tumor of the urinary system. The presence or absence of muscle invasion in early bladder cancer is an independent prognostic factor. The involvement of muscle invasion affects the choice of surgical methods and treatment. Preoperatively, the precise assessment of bladder cancer staging has important practical value. A more accurate preoperative assessment of bladder cancer staging can reduce overtreatment and provide a favorable basis for clinicians to choose more reasonable and effective surgical methods. Clinically, there has been a longstanding desire to diagnose the staging of bladder cancer through a simple, convenient, effective, and non-invasive examination. As relevant research progresses, a multi-omics diagnostic model will be beneficial in improving diagnostic efficiency. This project aims to establish a multi-omics artificial intelligence system based on deep learning and transfer learning to accurately diagnose the staging of bladder cancer and predict the efficacy of neoadjuvant chemotherapy. This system will assist in clinical treatment decision-making.

Research Content

1. Establishment of Ultrasound and Magnetic Resonance Imaging-Pathology-Clinical Dataset: This project aims to include medical records diagnosed through ultrasound and magnetic resonance imaging. Combining results from cytopathology or tissue pathology, the project will collect ultrasound images of the bladder, magnetic resonance images of the bladder, pathology images, and clinically relevant follow-up data. Tumor tissue specimens will be collected for immunohistochemical fluorescence staining. Participants with indistinct imaging findings or those with other primary malignant tumors will be excluded. The goal is to establish an ultrasound and magnetic resonance imaging-pathology-clinical dataset.

2. Establishment of an Artificial Intelligence System for Precise Diagnosis of Bladder Cancer Staging and Prognosis based on Ultrasound Imaging: Collecting clinical baseline information, surgical pathology, and ultrasound images of enrolled subjects. The ultrasound images will undergo homogenization processing. Utilizing algorithms based on U-Net, Transformer, and attention mechanisms for tissue segmentation and feature extraction, a deep learning model based on convolutional neural networks will be established for precise diagnosis of bladder cancer staging and prognosis:

   Construction of a mathematical model for staging bladder cancer using ultrasound contrast: Using convolutional neural networks for deep learning to build a mathematical model for staging bladder cancer. Developing an artificial intelligence diagnostic system for ultrasound contrast images based on deep learning and mathematical models to determine whether bladder cancer has muscle invasion.

   Construction of a mathematical model to discriminate prognosis features of bladder cancer using ultrasound imaging: Automatically delineating target areas and extracting ultrasound image features of bladder cancer lesions using new artificial intelligence technology - convolutional neural networks to build a model for evaluating the prognosis of bladder cancer lesions and achieving accurate prognosis diagnosis.

3. Establishment of an Artificial Intelligence System for the Joint Diagnosis of Bladder Cancer and Staging based on Ultrasound and Magnetic Resonance Imaging with Pathology: Based on bladder ultrasound, magnetic resonance imaging, and pathology image data, the baseline information of study subjects will be digitized. Ultrasound images, magnetic resonance images, and pathology images will undergo homogenization processing. Utilizing algorithms based on U-Net, Transformer, and attention mechanisms for segmentation and feature extraction of bladder ultrasound images, magnetic resonance images, and pathology images, a deep learning model based on convolutional neural networks will be established for the precise diagnosis of bladder cancer staging and prognosis:

   Construction of a mathematical model for joint pathology-based staging of bladder cancer using ultrasound and magnetic resonance imaging: Using convolutional neural networks for deep learning to build a mathematical model for staging bladder cancer. Based on the mathematical model, continuously optimizing algorithms, developing multi-omics, multidimensional artificial intelligence diagnostic systems based on ultrasound images, magnetic resonance images, pathology images, and clinical features, achieving accurate diagnosis of bladder cancer staging and prognosis prediction models.

   Construction of a mathematical model to discriminate prognosis features of bladder cancer using joint ultrasound, magnetic resonance, and pathology: Automatically delineating target areas and extracting features of bladder cancer lesions using convolutional neural networks and new artificial intelligence technology. Building a mathematical model to evaluate the prognosis of bladder cancer lesions.

   After the completion of the multi-omics, multidimensional artificial intelligence precise prediction model, validate the reliability of the model in prospective observational cohort study data and implement an intelligent system to assist in improving the efficiency of doctor diagnosis.

4. Establishment of an Artificial Intelligence System based on Dynamic Quantitative Immune Cell Maps using Ultrasound Imaging: Based on bladder ultrasound and pathological immunohistochemistry data, artificial intelligence algorithms will accurately determine the expression of immune cells through ultrasound images. Visualization of 2D spatial distribution heatmaps will be conducted to explore the intercorrelation features between ultrasound images and immune cells. Study subjects will be divided into training and validation cohorts. The baseline information of study subjects will be digitized. Ultrasound images will undergo homogenization processing. Utilizing algorithms based on U-Net, Transformer, attention mechanisms, and generative adversarial networks for tissue segmentation, feature extraction, and model establishment:

Construction of a mathematical model for quantitative immune cell maps using ultrasound images: Using new artificial intelligence technology - convolutional neural networks for deep learning to build a mathematical model for predicting the expression of immune cells. Developing an artificial intelligence diagnostic system for quantifying the microenvironment of ultrasound contrast images, determining the expression of immune cells in bladder cancer lesions.

The model will combine the immune cells with the 2D ultrasound image prediction heatmap, forming a visual 2D ultrasound image-immune cell heatmap. Exploring the spatial location of immune cells in ultrasound images. Using single-cell and spatial transcriptome sequencing methods to verify the accuracy of the spatial distribution of ultrasound image-quantified immune cells.

Inclusion Criteria:

1. Ultrasound and other imaging examinations (CT, MR, etc.) suggest bladder masses and are suspicious for bladder cancer patients.

2. The bladder is well filled, and no allergic reactions to ultrasound contrast agents are found.

3. No surgery or radiotherapy/chemotherapy has been performed.

4. Patients who meet the indications for surgical resection and are planned for surgical treatment, including one of the following:

   1. Clinical symptoms consistent with suspected bladder cancer (such as gross hematuria, etc.);
   2. Patients with confirmed primary or recurrent bladder cancer by cystoscopic biopsy;
   3. Rapid urine cytology and urine cytology FISH testing suggest malignancy.

Exclusion Criteria:

1. Individuals unable to tolerate surgery;
2. Individuals allergic to ultrasound contrast agents, unable to undergo ultrasound contrast examination;
3. Unsuccessful preoperative ultrasound contrast examination or non-compliant patients;
4. Postoperative pathology does not indicate bladder cancer;
5. Patients who have undergone chemotherapy or radiation therapy.