Paola Campadelli

Quantitative evaluation of color image segmentation results

In this paper we consider the problem of the automatic evaluation of the results of color image segmentation. Liu and Yang (1994) have proposed an evaluation function, inspired by the qualitative criteria for good image segmentation established by Haralick and Shapiro (1985), that does not require that the user set any parameter or threshold value. We identify some limitations in this evaluation function, and propose two enhanced functions that correspond more closely to visual judgment.

Liver segmentation from computed tomography scans: A survey and a new algorithm

OBJECTIVE In the recent years liver segmentation from computed tomography scans has gained a lot of importance in the field of medical image processing since it is the first and fundamental step of any automated technique for the automatic liver disease diagnosis, liver volume measurement, and 3D liver volume rendering. METHODS In this paper we report a review study about the semi-automatic and automatic liver segmentation techniques, and we describe our fully automatized method. RESULTS The survey reveals that automatic liver segmentation is still an open problem since various weaknesses and drawbacks of the proposed works must still be addressed. Our gray-level based liver segmentation method has been developed to tackle all these problems; when tested on 40 patients it achieves satisfactory results, comparable to the mean intra- and inter-observer variation. CONCLUSIONS We believe that our technique outperforms those presented in the literature; nevertheless, a common test set with its gold standard traced by experts, and a generally accepted performance measure are required to demonstrate it.

Neural encoding of input transients investigated by intracellular injection of ramp currents in cat α‐motoneurones

1. Input—output relations were analysed in spinal α‐motoneurones during current transients reaching a steady level after a linear growth of different slopes. The motoneurone output considered in the analysis was the instantaneous frequency of the cell discharge.

Precise eye localization through a general-to-specific model definition

We present a method for precise eye localization that uses two Support Vector Machines trained on properly selected Haar wavelet coefficients. The evaluation of our technique on many standard databases exhibits very good performance. Furthermore, we study the strong correlation between the eye localization error and the face recognition rate.

A face recognition system based on automatically determined facial fiducial points

In this paper, a completely automatic face recognition system is presented. The method works on color images: after having localized the face and the facial features, it determines 24 facial fiducial points, and characterizes them applying a bank of Gabor filters which extract the peculiar texture around them (jets). Recognition is realized measuring the similarity between the different jets. The system is inspired by the elastic bunch graph method, while it does no assumption on the scale, pose, and the background. Comparison with standard algorithms is presented and discussed.

Color image segmentation using Hopfield networks

Color image segmentation is frequently based on pixel classification, either supervised or unsupervised, without taking into account spatial information. This may generate noisy results. One technique proposed to solve this problem is the use of Hopfield neural networks. In this paper, we present two segmentation algorithms for color image segmentation based on Huangs idea of describing the segmentation problem as one of minimizing a suitable energy function for a Hopfield network. The first algorithm, which resembles Huangs algorithm for grey-level images, builds three different networks (one for each color feature considered), and then combines the results. The second builds a single network according to the number of clusters obtained by histogram analysis. We have changed the network initialization, its dynamic evolution, and the technique of histogram analysis employed in both with respect to the original proposition. The experimental results, heuristically and quantitatively evaluated, are encouraging.

PRECISE EYE AND MOUTH LOCALIZATION

The literature on the topic has shown a strong correlation between the degree of precision of face localization and the face recognition performance. Hence, there is a need for precise facial featu...

A Boosting Algorithm for Regression

A new boosting algorithm ADABOOST-Ra for regression problems is presented and upper bound on the error is obtained. Experimental results to compare ADABOOST-RΔ and other learning algorithms are given.

The dynamic response of cat gastrocnemius motor units investigated by ramp-current injection into their motoneurones.

1. The isometric force developed by single motor units in response to injection of ramp‐and‐hold currents into their motoneurones was recorded from the common tendon of the gastrocnemius muscles of the cat. The average rate of rise of the force (force‐slope) produced by the ramp‐evoked discharge, was found to grow almost linearly with the rate of current injection (current‐slope) up to a saturation value (maximal force‐slope). 2. The slope of the function which links the force slope to the current‐slope is the gain (dF/dI) of the motor unit under dynamic conditions. The value of the dynamic gain, measured in the linear region of growth, displays a large variability, i.e. for each nanoampere of current injected, the force developed is as much as 40 times larger in the strongest than in the weakest motor units. Such large gain differences, however, are drastically reduced if the force is expressed as a percentage of the maximal tetanic tension, Ft: per nanoampere injected, most of the units deliver from 1.0 to 3.0% of Ft. 3. The maximal force‐slope which each unit could reach exhibits a large variability, ranging from 0.06 to 4.0 g ms‐1. Like the dynamic gain, the maximal force‐slope is positively related to Ft. 4. It was found that the dynamic sensitivity of the motoneurone, i.e. the increase of the firing rate per unitary increase of the current‐slope, governs the fractional growth of the force‐slope, whereas the motor unit contraction time determines the firing rate at which maximal force‐slope is reached. Together, the two factors co‐operate in defining, for each motor unit, the range of input‐slopes within which the force‐slope is regulated. 5. The motoneurones which supply the weak motor units, those with the lowest dynamic gain, have higher dynamic sensitivity and lower rheobase than those innervating the strong motor units. This suggests that weak motor units need less synaptic current both to be recruited and to reach the maximal speed of force development when their input is supraliminal.

A segmentation framework for abdominal organs from CT scans

OBJECTIVE Computed tomography images are becoming an invaluable mean for abdominal organ investigation. In the field of medical image processing, some of the current interests are the automatic diagnosis of liver, spleen, and kidney pathologies, and the 3D volume rendering of these abdominal organs. Their automatic segmentation is the first and fundamental step in all these studies, but it is still an open problem. METHODS In this paper we propose a fully automatic, gray-level based segmentation framework based on a multiplanar fast marching method. The proposed segmentation scheme is general, and employs only established and not critical anatomical knowledge. For this reason, it can be easily adapted to segment different abdominal organs, by overcoming problems due to the high inter- and intra-patient gray-level, and shape variabilities; the extracted volumes are then combined to produce the final results. RESULTS The system has been evaluated by computing the symmetric volume overlap (SVO) between the automatically segmented (liver and spleen) volumes and the volumes manually traced by radiological experts. The test dataset is composed of 60 images, where 40 images belong to a private dataset, and 20 images to a public one. Liver segmentation has achieved an average SVO congruent with94, which is comparable to the mean intra- and inter-personal variation (96). Spleen segmentation achieves similar, promising results (SVO congruent with93). The comparison of these results with those achieved by active contour models (SVO congruent with90), and topology adaptive snakes (SVO congruent with92) proves the efficacy of our system. CONCLUSIONS The described segmentation method is a general framework that can be adapted to segment different abdominal organs, achieving promising segmentation results. It has to be noted that its performance could be further improved by incorporating shape based rules.