It has been shown that employing multiple atlas images improves segmentation accuracy in atlas-based medical image segmentation. Each atlas image is registered to the target image independently and the calculated transformation is applied to the segmentation of the atlas image to obtain a segmented version of the target image. Several independent candidate segmentations result from the process, which must be somehow combined into a single final segmentation. Majority voting is the generally used rule to fuse the segmentations, but more sophisticated methods have also been proposed. In this paper, we show that the use of global weights to ponderate candidate segmentations has a major limitation. As a means to improve segmentation accuracy, we propose the generalized local weighting voting method. Namely, the fusion weights adapt voxel-by-voxel according to a local estimation of segmentation performance. Using digital phantoms and MR images of the human brain, we demonstrate that the performance of each combination technique depends on the gray level contrast characteristics of the segmented region, and that no fusion method yields better results than the others for all the regions. In particular, we show that local combination strategies outperform global methods in segmenting high-contrast structures, while global techniques are less sensitive to noise when contrast between neighboring structures is low. We conclude that, in order to achieve the highest overall segmentation accuracy, the best combination method for each particular structure must be selected.

Histomorphometry is the most sensitive technique since it detects airspace enlargement before the other methods (1 h after treatment). Micro-CT correlates well with histology (r2 = 0.63) proving appropriate for longitudinal studies. Functional test parameters do not necessarily correlate with the extent of emphysema, as they can be influenced by acute inflammation. Finally, cytokine measurements correlate with the presence of inflammation in histology but not with emphysema.

Objective. To define the sensitivity of microcomputed tomography- (micro-CT-) derived descriptors for the quantification of lung damage caused by elastase instillation. Materials and Methods. The lungs of 30 elastase treated and 30 control A/J mice were analyzed 1, 6, 12, and 24 hours and 7 and 17 days after elastase instillation using (i) breath-hold-gated micro-CT, (ii) pulmonary function tests (PFTs), (iii) RT-PCR for RNA cytokine expression, and (iv) histomorphometry. For the latter, an automatic, parallel software toolset was implemented that computes the airspace enlargement descriptors: mean linear intercept (L m) and weighted means of airspace diameters (D 0, D 1, and D 2). A Support Vector Classifier was trained and tested based on three nonhistological descriptors using D 2 as ground truth. Results. D 2 detected statistically significant differences (P < 0.01) between the groups at all time points. Furthermore, D 2 at 1 hour (24 hours) was significantly lower (P < 0.01) than D 2 at 24 hours (7 days). The classifier trained on the micro-CT-derived descriptors achieves an area under the curve (AUC) of 0.95 well above the others (PFTS AUC = 0.71; cytokine AUC = 0.88). Conclusion. Micro-CT-derived descriptors are more sensitive than the other methods compared, to detect in vivo early signs of the disease.

Animal models of lung disease are gaining importance in understanding the underlying mechanisms of diseases such as emphysema and lung cancer. Micro-CT allows in vivo imaging of these models, thus permitting the study of the progression of the disease or the effect of therapeutic drugs in longitudinal studies. Automated analysis of micro-CT images can be helpful to understand the physiology of diseased lungs, especially when combined with measurements of respiratory system input impedance. In this work, we present a fast and robust murine airway segmentation and reconstruction algorithm. The algorithm is based on a propagating fast marching wavefront that, as it grows, divides the tree into segments. We devised a number of specific rules to guarantee that the front propagates only inside the airways and to avoid leaking into the parenchyma. The algorithm was tested on normal mice, a mouse model of chronic inflammation and a mouse model of emphysema. A comparison with manual segmentations of two independent observers shows that the specificity and sensitivity values of our method are comparable to the inter-observer variability, and radius measurements of the mainstem bronchi reveal significant differences between healthy and diseased mice. Combining measurements of the automatically segmented airways with the parameters of the constant phase model provides extra information on how disease affects lung function.

No abstract