SU-E-J-95: Towards Optimum Boundary Conditions for Biomechanical Model Based Deformable Registration Using Intensity...

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Computational Pathology

Background:

• Accurate registration of histology to in vivo MRI is vital for validating prostate cancer imaging.
• Geometric discrepancies between histology sections and in vivo MRI planes introduce errors in direct registration.
• Ex vivo high-resolution MRI acquisition can mitigate direct registration challenges.

Purpose of the Study:

• To develop and validate a novel intensity-based deformable registration algorithm for aligning histology to ex vivo MRI of prostatectomy specimens.
• To improve the accuracy of registration by incorporating intermediate images derived from local image intensity variations.
• To enhance the validation of in vivo prostate cancer imaging through precise histological-radiological correlation.

Main Methods:

• An intensity-based deformable registration algorithm was developed, utilizing standard deviation in local image windows to create intermediate MR and histology images.
• The Drop package was employed for initial registration of intermediate images, followed by a refinement step using a finer B-spline grid.
• Validation involved synthetic data using Magnitude of Differences (MOD) and Angular Error (AE), and clinical data using Target Registration Error (TRE) of manual landmarks.

Main Results:

• The proposed method using intermediate images achieved a MOD of 0.6mm and AE of 8.3 degrees, outperforming registration without intermediate images (0.8mm, 10.0 degrees).
• The average TRE for the proposed method was 1.0±0.6 mm, compared to 1.6±1.1 mm achieved with the Elastix software.
• These results demonstrate improved accuracy and robustness in deformable registration of prostatectomy specimens.

Conclusions:

• The developed intensity-based deformable registration algorithm, incorporating intermediate images, shows improved accuracy and robustness for prostatectomy specimens.
• This method offers a more precise tool for correlating histological findings with ex vivo MRI, aiding in the validation of in vivo prostate cancer imaging.
• The findings support the utility of this novel approach in advancing cancer imaging research and validation.