In Silico Oncology: Quantification of the In Vivo Antitumor Efficacy of Cisplatin-Based Doublet Therapy in Non-Small Cell Lung Cancer (NSCLC) through a Multiscale Mechanistic Model.

The 5-year survival of non-small cell lung cancer patients can be as low as 1% in advanced stages. For patients with resectable disease, the successful choice of preoperative chemotherapy is critical to eliminate micrometastasis and improve operability. In silico experimentations can suggest the optimal treatment protocol for each patient based on their own multiscale data. A determinant for reliable predictions is the a priori estimation of the drugs' cytotoxic efficacy on cancer cells for a given treatment. In the present work a mechanistic model of cancer response to treatment is applied for the estimation of a plausible value range of the cell killing efficacy of various cisplatin-based doublet regimens. Among others, the model incorporates the cancer related mechanism of uncontrolled proliferation, population heterogeneity, hypoxia and treatment resistance. The methodology is based on the provision of tumor volumetric data at two time points, before and after or during treatment. It takes into account the effect of tumor microenvironment and cell repopulation on treatment outcome. A thorough sensitivity analysis based on one-factor-at-a-time and latin hypercube sampling/partial rank correlation coefficient approaches has established the volume growth rate and the growth fraction at diagnosis as key features for more accurate estimates. The methodology is applied on the retrospective data of thirteen patients with non-small cell lung cancer who received cisplatin in combination with gemcitabine, vinorelbine or docetaxel in the neoadjuvant context. The selection of model input values has been guided by a comprehensive literature survey on cancer-specific proliferation kinetics. The latin hypercube sampling has been recruited to compensate for patient-specific uncertainties. Concluding, the present work provides a quantitative framework for the estimation of the in-vivo cell-killing ability of various chemotherapies. Correlation studies of such estimates with the molecular profile of patients could serve as a basis for reliable personalized predictions.

In silico oncology: exploiting clinical studies to clinically adapt and validate multiscale oncosimulators.

This paper presents a brief outline of the notion and the system of oncosimulator in conjunction with a high level description of the basics of its core multiscale model simulating clinical tumor response to treatment. The exemplary case of lung cancer preoperatively treated with a combination of chemotherapeutic agents is considered. The core oncosimulator model is based on a primarily top-down, discrete entity - discrete event multiscale simulation approach. The critical process of clinical adaptation of the model by exploiting sets of multiscale data originating from clinical studies/trials is also outlined. Concrete clinical adaptation results are presented. The adaptation process also conveys important aspects of the planned clinical validation procedure since the same type of multiscale data - although not the same data itself- is to be used for clinical validation. By having exploited actual clinical data in conjunction with plausible literature-based values of certain model parameters, a realistic tumor dynamics behavior has been demonstrated. The latter supports the potential of the specific oncosimulator to serve as a personalized treatment optimizer following an eventually successful completion of the clinical adaptation and validation process.

Automated model-based vertebra detection, identification, and segmentation in CT images.

For many orthopaedic, neurological, and oncological applications, an exact segmentation of the vertebral column including an identification of each vertebra is essential. However, although bony structures show high contrast in CT images, the segmentation and labelling of individual vertebrae is challenging. In this paper, we present a comprehensive solution for automatically detecting, identifying, and segmenting vertebrae in CT images. A framework has been designed that takes an arbitrary CT image, e.g., head-neck, thorax, lumbar, or whole spine, as input and provides a segmentation in form of labelled triangulated vertebra surface models. In order to obtain a robust processing chain, profound prior knowledge is applied through the use of various kinds of models covering shape, gradient, and appearance information. The framework has been tested on 64 CT images even including pathologies. In 56 cases, it was successfully applied resulting in a final mean point-to-surface segmentation error of 1.12+/-1.04mm. One key issue is a reliable identification of vertebrae. For a single vertebra, we achieve an identification success of more than 70%. Increasing the number of available vertebrae leads to an increase in the identification rate reaching 100% if 16 or more vertebrae are shown in the image.

Spine segmentation using articulated shape models.

Including prior shape in the form of anatomical models is a well-known approach for improving segmentation results in medical images. Currently, most approaches are focused on the modeling and segmentation of individual objects. In case of object constellations, a simultaneous segmentation of the ensemble that uses not only prior knowledge of individual shapes but also additional information about spatial relations between the objects is often beneficial. In this paper, we present a two-scale framework for the modeling and segmentation of the spine as an example for object constellations. The global spine shape is expressed as a consecution of local vertebra coordinate systems while individual vertebrae are modeled as triangulated surface meshes. Adaptation is performed by attracting the model to image features but restricting the attraction to a former learned shape. With the developed approach, we obtained a segmentation accuracy of 1.0 mm in average for ten thoracic CT images improving former results.

Structure of best possible strategies for finding ground states.

Finding the ground state of a system with a complex energy landscape is important for many physical problems including protein folding, spin glasses, chemical clusters, and neural networks. Such problems are usually solved by heuristic search methods whose efficacy is judged by empirical performance on selected examples. We present a proof that for a wide range of objective functions threshold accepting is the best possible strategy within a large class of algorithms that simulate random walks on the landscape. In particular, it can perform better than simulated annealing, Tsallis and Glauber statistics.