ABSTRACT
OBJECTIVES: To evaluate the ex vivo accuracy of an MRI-TRUS fusion device for guiding targeted prostate biopsies, to identify the origin of errors, and to evaluate the likelihood that lesions can be accurately targeted. MATERIALS AND METHODS: Three prostate phantoms were used to perform 27 biopsies using transperineal MRI-TRUS fusion. All phantoms underwent 3-T MRI. The prostate contour and nine lesions were delineated onto the MRI. A 3D-US dataset was generated and fused with the MRI. Per lesion, one needle was virtually planned. The postbiopsy needle location was virtually registered. The needle trajectory was marked using an MRI-safe guidewire. Postinterventional MRI was performed. The coordinates of the lesion on preinterventional MRI, the virtually planned needle, the virtually registered needle, and the marked needle trajectory on postinterventional MRI were documented and used to calculate the planning error (PE), targeting error (TE), and overall error (OE). Using the OE in the transversal plane, an upper one-sided tolerance interval was calculated to assess the likelihood that a biopsy needle was on target. RESULTS: In the transversal plane, the mean PE, TE, and OE were 1.18, 0.39, and 2.33 mm, respectively. Using a single biopsy core, the likelihood that lesions with a diameter of 2 mm can be accurately targeted is 26%; lesions of 3 mm 61%; lesions of 4 mm 86%; lesions of 5 mm 96%; and lesions of 6 mm 99%. The likelihood of accurate sampling increases if more biopsy cores are used. CONCLUSION: MRI-TRUS fusion allows for accurate sampling of MRI-identified lesions with an OE of 2.33 mm. Lesions with a diameter of 3 mm or more can be accurately targeted. These results should be considered the lower limit of in vivo accuracy.
