Improved Cone Beam CT Methods for Intraoperative Resection Margin Determination in Breast Cancer Surgery

DFG project no. 534069778

The original aim of this DFG-funded project is to improve the intraoperative assessment of breast cancer resection margins using high-resolution X-ray imaging. A particular focus is the reliable detection and localization of microcalcifications in surgical breast specimens, which can provide important information about the extent of malignant disease.

To address this challenge, the project combines advanced photon-counting detectors with optimized acquisition and reconstruction methods. An important component of this work is the development of Photon-Counting Super-Sampled Imaging (PC-SSI), a motion-based imaging approach designed to overcome the limitations imposed by the native pixel size of photon-counting detectors.

In PC-SSI, the detector is deliberately moved during image acquisition so that the object is sampled at multiple sub-pixel positions. These measurements are registered with sub-pixel precision and combined using a physics-based iterative reconstruction approach. The original project therefore provides not only an optimized imaging approach for breast specimens, but also a platform for investigating the limits of spatial resolution achievable with photon-counting detectors.

From a clinical question to a new imaging technology

Our experimental work demonstrated that mechanical super-sampling can substantially improve the effective spatial resolution of photon-counting radiography. In combination with model-based iterative reconstruction, we achieved an approximately three-fold improvement in measured spatial resolution, while also reducing the influence of detector inhomogeneities and other hardware-related limitations.

These results were particularly encouraging because they demonstrated that the approach is not limited by the nominal pixel pitch of the detector. Instead, detector motion can be used as an additional sampling dimension, allowing information below the native pixel size to be recovered.

Towards high-resolution mammography

The promising performance of PC-SSI naturally raises the question of whether the same approach can be used beyond specimen imaging.

Mammography is an especially attractive application. The detection of very small microcalcifications is already an established strength of mammographic imaging, but their detailed morphology can be difficult to resolve. Higher spatial resolution could provide additional information about the size, shape and distribution of these structures.

We therefore began investigating the application of photon-counting super-resolution imaging to mammography. Experiments with a mammographic accreditation phantom demonstrated improved contrast and signal-to-noise characteristics compared with a clinical mammography system. At the same time, the super-sampling process helped to average out local detector response variations, improving image homogeneity.

These results provide the basis for our ongoing efforts to translate PC-SSI from experimental high-resolution imaging into a clinically relevant mammographic imaging technology.

Looking beyond detection: morphology and spectral information

The potential of super-resolution mammography goes beyond simply detecting smaller structures.

Microcalcifications are characterized not only by their presence, but also by their morphology, distribution and composition. With improved spatial resolution and the spectral capabilities of photon-counting detectors, we aim to investigate whether these additional features can provide information related to the underlying pathology.

Our current work therefore combines high-resolution photon-counting imaging with spectral analysis, histopathological reference data and computational methods. This creates a pathway from the physical structure observed in an image to its biological interpretation.