研究目的
To develop a noninvasive and accurate diagnostic imaging method for bladder cancer detection using nanoscale-resolution scanning of cell surfaces from body fluids and machine-learning analysis.
研究成果
The presented method demonstrates a statistically significant improvement in diagnosing bladder cancer with 94% diagnostic accuracy when examining five cells per patient’s urine sample. It offers a noninvasive, accurate, and potentially cost-effective alternative to current clinical standards.
研究不足
The method requires further validation with larger cohorts of patients before clinical introduction. The accuracy may vary with the number of cells analyzed per patient.
1：Experimental Design and Method Selection:
Utilized subresonance tapping AFM modalities (PeakForce and Ringing mode) for nanoscale-resolution imaging of cell surfaces. Machine-learning methods (Random Forest, Extremely Randomized Forest, Gradient Boosting Trees) were applied to analyze the images.
2：Sample Selection and Data Sources:
Urine samples from 43 individuals without evidence of bladder cancer and 25 cancer patients with pathologically confirmed bladder cancer were analyzed.
3：List of Experimental Equipment and Materials:
Atomic force microscopy (AFM) for imaging, optical microscope for cell location, and machine-learning algorithms for data analysis.
4：Experimental Procedures and Operational Workflow:
Cells were collected from urine, fixed, washed, freeze-dried on a glass slide, and imaged using AFM. Machine-learning analysis was performed on the images to classify cells.
5：Data Analysis Methods:
Machine-learning methods were used to analyze surface parameters derived from AFM images. The accuracy of cancer detection was evaluated based on the analysis of one to five cells per patient.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
