Mieke Bus

92 Chapter 6 Results Specimen characteristics are listed in table 2. All the patients underwent a diagnostic URS including biopsies before radical nephroureterectomy. None of the patients received neo-adjuvant therapy. Table 2: Patient Characteristics N Gender Age (yr) Tumor Stage Tumor Grade Tumor Location 1 F 84 T1 2 (high grade) renal pelvis 2 F 48 Ta 2 (low grade) distal ureter 3 F 65 T1 3 renal pelvis 4 M 74 T2 3 renal pelvis 5 M 76 T1 3 distal and proximal ureter, renal pelvis CT based co-registration CT based co-registration of ELUS and OCT data was achieved for 5 patients. Figure 2 depicts an exemplary semitransparent 3D rendering of the CT dataset which was used for co-local- ization of the OCT and CT datasets. Within the CT rendering, a non-transparent purple probe has been made visible. The purple probe was automatically segmented based on Hounsfield units. The curvature of the probe was used to manually deform the OCT and ELUS datasets as shown in Figure 3. A longitu- dinal cross-section of both the OCT and ELUS dataset shows a clear difference in imaging depth. Additional visible markers, such as the indentation caused by surgical clip that was placed intra operatively and was removed before introducing the imaging probes show the accurateness of dataset matching. OCT and ELUS imaging of the upper urinary tract In OCT images of normal appearing renal pelvis and ureter, the urothelium, lamina pro- pria and muscularis propria were clearly visible. In ELUS images of normal appearing ureter, anatomical layers could be distinguished, although the resolution was lower compared to OCT images and because of this low-resolution imaging, the urothelial layer could not be identified (table 3). In OCT images with visible lesions, the anatomical layers could not be identified in all OCT datasets, resulting in a total identification of the urothelium in 79.4%, lamina propria in 79.4%, muscularis propria in 82.4% and periureteral fat in 50% of the total OCT datasets (table 3). In ELUS images the inner mucosal layer is seen as a hyperechoic layer and recognized in 80.9% of the ELUS datasets. The muscularis layer is hypoechoic and seen in 80.9% of the ELUS datasets and the periureteric fat hyperechoic and recognized in 80.9% of the ELUS datasets.