Distance measures for image segmentation evaluation

The task considered in this paper is performance evaluation of region segmentation algorithms in the ground-truth-based paradigm. Given a machine segmentation and a ground-truth segmentation, performance measures are needed. We propose to consider the image segmentation problem as one of data clustering and, as a consequence, to use measures for comparing clusterings developed in statistics and machine learning. By doing so, we obtain a variety of performance measures which have not been used before in image processing. In particular, some of these measures have the highly desired property of being a metric. Experimental results are reported on both synthetic and real data to validate the measures and compare them with others.

Dissociation between overt and unconscious face processing in fusiform face area

The precise role of the fusiform face area (FFA) in face processing remains controversial. In this study, we investigated to what degree FFA activation reflects additional functions beyond face perception. Seven volunteers underwent rapid event-related functional magnetic resonance imaging while they performed a face-encoding and a face-recognition task. During face encoding, activity in the FFA for individual faces predicted whether the individual face was subsequently remembered or forgotten. However, during face recognition, no difference in FFA activity between consciously remembered and forgotten faces was observed, but the activity of FFA differentiated if a face had been seen previously or not. This demonstrated a dissociation between overt recognition and unconscious discrimination of stimuli, suggesting that physiological processes of face recognition can take place, even if not all of its operations are made available to consciousness.

Functional imaging of visuospatial processing in Alzheimer's disease

Alzheimer's disease (AD) is known to cause a variety of disturbances of higher visual functions that are closely related to the neuropathological changes. Visual association areas are more affected than primary visual cortex. Additionally, there is evidence from neuropsychological and imaging studies during rest or passive visual stimulation that the occipitotemporal pathway is less affected than the parietal pathway. Our goal was to investigate functional activation patterns during active visuospatial processing in AD patients and the impact of local cerebral atrophy on the strength of functional activation. Fourteen AD patients and fourteen age-matched controls were measured with functional magnetic resonance imaging (fMRI) while they performed an angle discrimination task. Both groups revealed overlapping networks engaged in angle discrimination including the superior parietal lobule (SPL), frontal and occipitotemporal (OTC) cortical regions, primary visual cortex, basal ganglia, and thalamus. The most pronounced differences between the two groups were found in the SPL (more activity in controls) and OTC (more activity in patients). The differences in functional activation between the AD patients and controls were partly explained by the differences in individual SPL atrophy. These results indicate that parietal dysfunction in mild to moderate AD is compensated by recruitment of the ventral visual pathway. We furthermore suggest that local cerebral atrophy should be considered as a covariate in functional imaging studies of neurodegenerative disorders.

The tragedy of the commons in evolutionary biology

Garrett Hardin's tragedy of the commons is an analogy that shows how individuals driven by self-interest can end up destroying the resource upon which they all depend. The proposed solutions for humans rely on highly advanced skills such as negotiation, which raises the question of how non-human organisms manage to resolve similar tragedies. In recent years, this question has promoted evolutionary biologists to apply the tragedy of the commons to a wide range of biological systems. Here, we provide tools to categorize different types of tragedy and review different mechanisms, including kinship, policing and diminishing returns that can resolve conflicts that could otherwise end in tragedy. A central open question, however, is how often biological systems are able to resolve these scenarios rather than drive themselves extinct through individual-level selection favouring self-interested behaviours.

Effect of X-ray tube parameters, iodine concentration, and patient size on image quality in pulmonary computed tomography angiography: a chest-phantom-study

OBJECTIVES: The aim of this phantom study was to evaluate the contrast-to-noise ratio (CNR) in pulmonary computed tomography (CT)-angiography for 300 and 400 mg iodine/mL contrast media using variable x-ray tube parameters and patient sizes. We also analyzed the possible strategies of dose reduction in patients with different sizes. MATERIALS AND METHODS: The segmental pulmonary arteries were simulated by plastic tubes filled with 1:30 diluted solutions of 300 and 400 mg iodine/mL contrast media in a chest phantom mimicking thick, intermediate, and thin patients. Volume scanning was done with a CT scanner at 80, 100, 120, and 140 kVp. Tube current-time products (mAs) varied between 50 and 120% of the optimal value given by the built-in automatic dose optimization protocol. Attenuation values and CNR for both contrast media were evaluated and compared with the volume CT dose index (CTDI(vol)). Figure of merit, calculated as CNR/CTDIvol, was used to quantify image quality improvement per exposure risk to the patient. RESULTS: Attenuation of iodinated contrast media increased both with decreasing tube voltage and patient size. A CTDIvol reduction by 44% was achieved in the thin phantom with the use of 80 instead of 140 kVp without deterioration of CNR. Figure of merit correlated with kVp in the thin phantom (r = -0.897 to -0.999; P < 0.05) but not in the intermediate and thick phantoms (P = 0.09-0.71), reflecting a decreasing benefit of tube voltage reduction on image quality as the thickness of the phantom increased. Compared with the 300 mg iodine/mL concentration, the same CNR for 400 mg iodine/mL contrast medium was achieved at a lower CTDIvol by 18 to 40%, depending on phantom size and applied tube voltage. CONCLUSIONS: Low kVp protocols for pulmonary embolism are potentially advantageous especially in thin and, to a lesser extent, in intermediate patients. Thin patients profit from low voltage protocols preserving a good CNR at a lower exposure. The use of 80 kVp in obese patients may be problematic because of the limitation of the tube current available, reduced CNR, and high skin dose. The high CNR of the 400 mg iodine/mL contrast medium together with lower tube energy and/or current can be used for exposure reduction.