Real-time brain oscillation detection and phase-locked stimulation using autoregressive spectral estimation and time-series forward prediction.

Neural oscillations are important features in a working central nervous system, facilitating efficient communication across large networks of neurons. They are implicated in a diverse range of processes such as synchronization and synaptic plasticity, and can be seen in a variety of cognitive processes. For example, hippocampal theta oscillations are thought to be a crucial component of memory encoding and retrieval. To better study the role of these oscillations in various cognitive processes, and to be able to build clinical applications around them, accurate and precise estimations of the instantaneous frequency and phase are required. Here, we present methodology based on autoregressive modeling to accomplish this in real time. This allows the targeting of stimulation to a specific phase of a detected oscillation. We first assess performance of the algorithm on two signals where the exact phase and frequency are known. Then, using intracranial EEG recorded from two patients performing a Sternberg memory task, we characterize our algorithm's phase-locking performance on physiologic theta oscillations: optimizing algorithm parameters on the first patient using a genetic algorithm, we carried out cross-validation procedures on subsequent trials and electrodes within the same patient, as well as on data recorded from the second patient.

Improved web-based calculators for predicting breast carcinoma outcomes.

We describe a set of web-based calculators, available at http://www.CancerMath.net , which estimate the risk of breast carcinoma death, the reduction in life expectancy, and the impact of various adjuvant treatment choices. The published SNAP method of the binary biological model of cancer metastasis uses information on tumor size, nodal status, and other prognostic factors to accurately estimate of breast cancer lethality at 15 years after diagnosis. By combining these 15-year lethality estimates with data on the breast cancer hazard function, breast cancer lethality can be estimated at each of the 15 years after diagnosis. A web-based calculator was then created to visualize the estimated lethality with and without a range of adjuvant therapy options at any of the 15 years after diagnosis, and enable conditional survival calculations. NIH population data was used to estimate non-breast-cancer chance of death. The accuracy of the calculators was tested against two large breast carcinoma datasets: 7,907 patients seen at two academic hospitals and 362,491 patients from the SEER national dataset. The calculators were found to be highly accurate and specific, as seen by their capacity for stratifying patients into groups differing by as little as a 2% risk of death, and accurately accounting for nodal status, histology, grade, age, and hormone receptor status. Our breast carcinoma calculators provide accurate and useful estimates of the risk of death, which can aid in analysis of the various adjuvant therapy options available to each patient.

A method for real-time cortical oscillation detection and phase-locked stimulation.

Neural oscillations are important features in a working central nervous system, facilitating efficient communication across large networks of neurons. To better study the role of these oscillations in various cognitive processes, and to be able to build clinical applications around them, accurate and precise estimations of the instantaneous frequency and phase are required. Here, we present methodology based on autoregressive modeling to accomplish this in real time. This allows the targeting of stimulation to a specific phase of a detected oscillation. Using intracranial EEG recorded from two patients performing a Sternberg memory task, we characterize our algorithm's phase-locking performance on physiologic theta oscillations.

An agent-based model identifies MRI regions of probable tumor invasion in a patient with glioblastoma.

We present an application of a previously developed agent-based glioma model (Chen et al 2009 Biosystems 95 234-42) for predicting spatio-temporal tumor progression using a patient-specific MRI lattice derived from apparent diffusion coefficient (ADC) data. Agents representing collections of migrating glioma cells are initialized based upon voxels at the outer border of the tumor identified on T1-weighted (Gd+) MRI at an initial time point. These simulated migratory cells exhibit a specific biologically inspired spatial search paradigm, representing a weighting of the differential contribution from haptotactic permission and biomechanical resistance on the migration decision process. ADC data from 9 months after the initial tumor resection were used to select the best search paradigm for the simulation, which was initiated using data from 6 months after the initial operation. Using this search paradigm, 100 simulations were performed to derive a probabilistic map of tumor invasion locations. The simulation was able to successfully predict a recurrence in the dorsal/posterior aspect long before it was depicted on T1-weighted MRI, 18 months after the initial operation.

Occult nipple involvement in breast cancer: clinicopathologic findings in 316 consecutive mastectomy specimens.

PURPOSE: Although breast-conserving surgery is a standard approach for patients with breast cancer, mastectomy often becomes necessary. Surgical options now include nipple-sparing mastectomy but its oncological safety is still controversial. This study evaluates frequency and patterns of occult nipple involvement in a large contemporary cohort of patients with the retroareolar margin as possible indicator of nipple involvement. PATIENTS AND METHODS: Three hundred sixteen consecutive mastectomy specimens (232 therapeutic, 84 prophylactic) with grossly unremarkable nipples were evaluated by coronal sections through the entire nipple and subareolar tissue. Extent and location of nipple involvement by carcinoma was assessed with the tissue deep to the skin as potential retroareolar en-face resection margin. RESULTS: Seventy-one percent of nipples from therapeutic mastectomies showed no pathologic abnormality, 21% had ductal carcinoma in situ (DCIS), invasive carcinoma (IC), or lymphovascular invasion (LVI), and 8% lobular neoplasia (lobular carcinoma in situ). Human epidermal growth factor receptor 2 amplification, tumor size, and tumor-nipple distance were associated with nipple involvement by multivariate analysis (P = .0047, .0126, and .0176); histologic grade of both DCIS (P = .002) and IC (P = .03), LVI (P = .03), and lymph node involvement (P = .02) by univariate analysis. Nipple involvement by IC or DCIS was identified in the retroareolar margin with a sensitivity of 0.8 and a negative predictive value of 0.96. None of the 84 prophylactic mastectomies showed nipple involvement by IC or DCIS. CONCLUSION: Nipple-sparing mastectomy may be suitable for selected cases of breast carcinoma with low probability of nipple involvement by carcinoma and prophylactic procedures. A retroareolar en-face margin may be used to test for occult involvement in patients undergoing nipple-sparing mastectomy.

Why cancer at the primary site and in the lymph nodes contributes to the risk of cancer death.

BACKGROUND: : Cancer at both the primary site and in the lymph nodes is associated with lethality, although the mechanism by which lethality arises from each site has been poorly understood. For breast carcinoma, each positive lymph node contributes an approximately 6% risk of death, and each millimeter of primary tumor greatest dimension contributes approximately 1%; whereas, for melanoma, each positive lymph node contributes an approximately 23% risk, and each millimeter of tumor thickness contributes approximately 8%: This is described by a pair of linked equations, the Size+Nodes method. METHODS: : A simple expression, the ProbabilityEstimation equation, which was derived from the authors' binary-biologic model of cancer metastasis, was used to calculate the probabilities of spread of cancer cells from data on tumor size, lymph node status, and death rate. RESULTS: : In this report, the authors demonstrated, that when similar masses of cancer are compared, the chance of lethal spread of a cancer cell to the periphery is approximately the same whether the spread emerges from a lymph node or from the primary site. The greater the number of cells at the primary site (tumor size) or the greater the number of cells in the lymph nodes (number of positive lymph nodes), the greater is the aggregate chance that 1 or more cells has undergone a lethal event of spread, a process captured by the Size+Nodes equations. CONCLUSIONS: : The lethal contributions of cancer at the primary site and lymph nodes can be explained by a simple mechanical process of the spread of cancer cells occurring with definable probabilities per cell. The presence of cancer in the lymph nodes does not indicate an intrinsic change in a malignancy but, rather, an increased mass of cancer from which spread can emerge. Cancer 2009. (c) 2009 American Cancer Society.

How cancer at the primary site and in the lymph nodes contributes to the risk of cancer death.

BACKGROUND: : It has long been appreciated that tumor size, lymph node status, and patient survival are related qualities, although how to isolate their interactions has not been obvious, nor has it been obvious how to integrate tumor size and lymph node status into predictions of the risk of death for individual patients. METHODS: : The authors describe a mathematical method, the binary-biological model of cancer metastasis, based on the spread of cancer cells, in which the equations capture the relations between tumor size, lymph node status, and cancer lethality. RESULTS: : For melanoma, renal cell carcinoma, and breast carcinoma, the relation between tumor size and the risk of cancer death was captured by the SizeOnly equation. For melanoma and breast carcinoma, the relation between tumor size and the presence of cancer in the lymph nodes was captured by using the NodalSizeOnly equation. For lymph node-negative melanoma and breast carcinoma, the relation between tumor size and risk of death was captured by the PrimarySizeOnly equation. For breast carcinoma, the model indicated that each positive lymph node contributed approximately 6% extra risk of death, whereas each millimeter of greatest primary tumor dimension contributed approximately 1% risk of death. For melanoma, each positive lymph node contributed approximately 23% risk of death, whereas each millimeter of primary melanoma thickness contributed approximately 8% risk of death. This information was captured by a pair of linked equations, the Size+Nodes method. CONCLUSIONS: : Both tumor size and the number of positive lymph nodes made independent contributions to the risk of cancer death, as estimated by using the Size+Nodes method. Cancer 2009. (c) 2009 American Cancer Society.

The impact of primary tumor size, lymph node status, and other prognostic factors on the risk of cancer death.

BACKGROUND: : Although many prognostic factors are associated with differences in cancer lethality, it may not be obvious whether a factor truly makes an independent contribution to lethality or simply is correlated with tumor size. There is currently no method for integrating tumor size, lymph node status, and other prognostic information from a patient into a single risk of death estimate. METHODS: : The SizeOnly equation, which captures the relation between tumor size and risk of death, makes it possible to determine whether a prognostic factor truly makes an independent contribution to cancer lethally or merely is associated with tumor size (SizeAssessment method). The magnitude of each factor's lethal contribution can be quantified by a parameter, g, inserted into the SizeOnly equation (PrognosticMeasurement method). A series of linked equations (the Size+Nodes+PrognosticFactors [SNAP] method) combines information on tumor size, lymph node status, and other prognostic factors from a patient into a single estimate of the risk of death. RESULTS: : Nine prognostic factors were identified that made marked, independent contributions to breast carcinoma lethality: grade; mucinous, medullary, tubular, and scirrhous adenocarcinoma; male sex; inflammatory disease; Paget disease; and lymph node status. In addition, it was determined that lymph node status made an independent contribution to melanoma lethality. The SNAP method was able to accurately estimate the risk of death and to finely stratify patients by risk. CONCLUSIONS: : The methods described provide a new framework for identifying and quantifying those factors that contribute to cancer lethality and provide a basis for web-based calculators (available at: http://www.CancerMath.net accessed July 29, 2009) that accurately estimate the risk of death for each patient. Cancer 2009. (c) 2009 American Cancer Society.

Cancer cell motility: optimizing spatial search strategies.

Aberrantly regulated cell motility is a hallmark of cancer cells. A hybrid agent-based model has been developed to investigate the synergistic and antagonistic cell motility-impacting effects of three microenvironment variables simultaneously: chemoattraction, haptotactic permission, and biomechanical constraint or resistance. Reflecting distinct cell-specific intracellular machinery, the cancer cells are modeled as processing a variety of spatial search strategies that respond to these three influencing factors with differential weights attached to each. While responding exclusively to chemoattraction optimizes cell displacement effectiveness, incorporating permission and resistance components becomes increasingly important with greater distance to the chemoattractant source and/or after reducing the ligand's effective diffusion coefficient. Extending this to a heterogeneous population of cells shows that displacement effectiveness increases with clonal diversity as characterized by the Shannon index. However, the resulting data can be fit best to an exponential function, suggesting that there is a level of population heterogeneity beyond which its added value to the cancer system becomes minimal as directionality ceases to increase. Possible experimental extensions and potential clinical implications are discussed.

Multi-scale, multi-resolution brain cancer modeling.

In advancing discrete-based computational cancer models towards clinical applications, one faces the dilemma of how to deal with an ever growing amount of biomedical data that ought to be incorporated eventually in one form or another. Model scalability becomes of paramount interest. In an effort to start addressing this critical issue, here, we present a novel multi-scale and multi-resolution agent-based in silico glioma model. While 'multi-scale' refers to employing an epidermal growth factor receptor (EGFR)-driven molecular network to process cellular phenotypic decisions within the micro-macroscopic environment, 'multi-resolution' is achieved through algorithms that classify cells to either active or inactive spatial clusters, which determine the resolution they are simulated at. The aim is to assign computational resources where and when they matter most for maintaining or improving the predictive power of the algorithm, onto specific tumor areas and at particular times. Using a previously described 2D brain tumor model, we have developed four different computational methods for achieving the multi-resolution scheme, three of which are designed to dynamically train on the high-resolution simulation that serves as control. To quantify the algorithms' performance, we rank them by weighing the distinct computational time savings of the simulation runs versus the methods' ability to accurately reproduce the high-resolution results of the control. Finally, to demonstrate the flexibility of the underlying concept, we show the added value of combining the two highest-ranked methods. The main finding of this work is that by pursuing a multi-resolution approach, one can reduce the computation time of a discrete-based model substantially while still maintaining a comparably high predictive power. This hints at even more computational savings in the more realistic 3D setting over time, and thus appears to outline a possible path to achieve scalability for the all-important clinical translation.