Electronic Thesis and Dissertation Repository


Master of Engineering Science


Biomedical Engineering


Dr. Aaron D Ward


An accurate assessment of prostate tumour burden supports appropriate treatment selection, ranging from active surveillance through focal therapy, to radical whole-prostate therapies. For selected patients, knowledge of the three-dimensional locations and sizes of prostate tumours on pre-procedural imaging supports planning of effective focal therapies that preferentially target tumours, while sparing surrounding healthy tissue. In the post-prostatectomy context, pathologic measurement of tumour burden in the surgical specimen may be an independent prognostic factor determining the need for potentially life-saving adjuvant therapy. An accurate and repeatable method for tumour volume assessment based on histology sections taken from the surgical specimen would be supportive both to the clinical workflow in the post-prostatectomy setting and to imaging validation studies correlating tumour burden measurements on pre-prostatectomy imaging with reference standard histologic tumour volume measurements.

Digital histopathology imaging is enabling a transition to a more objective quantification of some surgical pathology assessments, such as tumour volume, that are currently visually estimated by pathologists and subject to inter-observer variability. Histologic tumour volume measurement is challenged by the traditional 3–5 mm sparse spacing of images acquired from sections of radical prostatectomy specimens. Tumour volume estimates may benefit from a well-motivated approach to inter-slide tumour boundary interpolation that crosses these large gaps in a smooth fashion. This thesis describes a new level set-based shape interpolation method that reconstructs smooth 3D shapes based on arbitrary 2D tumour contours on digital histology slides. We measured the accuracy of this approach and used it as a reference standard against which to compare previous approaches in the literature that are simpler to implement in a clinical workflow, with the aim of determining a method for histologic tumour volume estimation that is both accurate and amenable to widespread implementation. We also measured the effect of decreasing inter-slide spacing on the repeatability of histologic tumour volume estimation. Furthermore, we used this histologic reference standard for tumour volume to measure the accuracy, inter-observer variability, and inter-sequence variability of prostate tumour volume estimation based on radiologists’ contouring of multi-parametric magnetic resonance imaging (MPMRI).

Our key findings were that (1) simple approaches to histologic tumour volume estimation that are based on 2- or 3-dimensional linear tumour measurements are more accurate than those based on 1-dimensional measurements; (2) although tumour shapes produced by smooth through-slide interpolation are qualitatively substantially different from those obtained from a planimetric approach normally used as a reference standard for histologic tumour volume, the volumes obtained were similar; (3) decreasing inter-slide spacing increases repeatability of histologic tumour volume estimates, and this repeatability decreases rapidly for inter-slide spacing values greater than 5 mm; (4) on MPMRI, observers consistently overestimated tumour volume as compared to the histologic reference standard; and (5) inter-sequence variability in MPMRI-based tumour volume estimation exceeded inter-observer variability.