- A transrectal ultrasound-guided prostate biopsy has a 20% false-negative rate.
- This 3D computational model of the prostate allows for mapping of regions suspected as cancerous onto an ultrasound image for precision prostate biopsy.
- Reduces the false-negative rate associated with the random sampling nature of TRUS-guided biopsy alone and reduces the need for repeat biopsies.
Prostate cancer is the most prevalent form of cancer in men. As of 2018, 3.24 million Americans were estimated to live with prostate cancer. This year alone an additional 268,490 new cases will be diagnosed. Given the high prevalence of prostate cancer, it is estimated that approximately 12.5% of men will be diagnosed with prostate cancer at some point in their life. Despite having a high five-year survival rate (97.5%), prostate cancer is currently the second leading cause of death in men.
Transrectal ultrasound (TRUS)-guided prostate biopsy is a standard prostate diagnostic procedure wherein a physician uses an ultrasound image to guide the excision of samples from the prostate. The number of prostate biopsies globally has reached 3.4 million with TRUS guided biopsies making up >95% of these procedures. Though TRUS is able to visualize the shape of the prostate, it cannot differentiate between normal and cancerous tissue inside the prostate. In light of this, clinicians take 10-14 biopsy samples based on a grid partitioning of the prostate which is applied to all patients unilaterally. This “one size fits all approach” leads to high false-negative rates (>20%) wherein samples were collected at sites distal to the growing tumor. Even if the cancerous tumor is correctly sampled and diagnosed, the tumor may be sampled incompletely leading to an incorrect determination of the stage of cancer. It is reported that 30-45% of prostate cancers diagnosed by TRUS guided biopsy are upstaged after radical prostatectomy.
Researchers at the Department of Radiation Oncology at The University of North Carolina at Chapel Hill have developed a tool for generating a 3D computational model of a patient’s prostate from 2D MRI images and mapping this model onto live imaging data acquired during TRUS guided biopsy. The 3D computational model was trained using images of prostates segmented by experts. The model was further refined by using statistical analysis of the shape and intensity variation and implementing the finite element method. This allowed the 3D computational model to capture information about the shape variation/deformation of the prostate during 2D image acquisition due to breathing or internal motion. The accuracy of the 3D computational model was tested on a data set consisting of 5 patients that were segmented by experts and found to have a Dice Similarity Coefficient of 92.1%. This computational 3D model can also be used to identify and map a target region of tissue suspicious of cancer.
- Lower false-positive rates than TRUS guided biopsy alone.
- Reduce the need for repeat biopsy
- More accurate cancer stage determination
- Greater computational model accuracy afforded by training on an MRI based image data set over an ultrasound imaging data set
A powerful application of this 3D computational model is the mapping to a live TRUS image of the prostate during TRUS guided biopsy. This allows the clinician to sample target regions suspicious of cancer with high precision and proper tumor sampling. This approach could dramatically reduce the false negative rates associated with the random sampling nature of TRUS-guided biopsy alone and reduce the need for repeat biopsy. This method also offers an advantage over ultrasound-generated deformable shape models for 3D prostate image segmentation because MRI imaging is better suited for visualizing prostate substructures than ultrasound imaging.
Name: Matthew Howe