Background The extraction and analysis of image features (radiomics) is a promising field in the precision medicine era,
with applications to prognosis, prediction, and response to treatment quantification. In this work, we present a mutual
information – based method for quantifying reproducibility of features, a necessary step for qualification before their
inclusion in big data systems.
Materials and Methods Ten patients with Non-Small Cell Lung Cancer (NSCLC) lesions were followed over time (7
time points in average) with Computed Tomography (CT). Five observers segmented lesions by using a semi-automatic
method and 27 features describing shape and intensity distribution were extracted. Inter-observer reproducibility was
assessed by computing the multi-information (MI) of feature changes over time, and the variability of global extrema.
Results The highest MI values were obtained for volume-based features (VBF). The lesion mass (M), surface to volume
ratio (SVR) and volume (V) presented statistically significant higher values of MI than the rest of features. Within the
same VBF group, SVR showed also the lowest variability of extrema. The correlation coefficient (CC) of feature values
was unable to make a difference between features.
Conclusions MI allowed to discriminate three features (M, SVR, and V) from the rest in a statistically significant manner.
This result is consistent with the order obtained when sorting features by increasing values of extrema variability. MI is a
promising alternative for selecting features to be considered as surrogate biomarkers in a precision medicine context.
The effect of atmospheric turbulence on the imaging of scenes, for a horizontal propagation of the light over a distance of 20 km, 15 meters above the sea surface, was analyzed at visible wavelength using a 20 cm telescope. Point-sources images were recorded during the night, and the fried parameter r<SUB>0</SUB> was derived leading to values ranging from 1.5 to 3.6 cm. A very high level of scintillation was observed. Studies of correlations between close-by sources lead to a very small domain of isoplanatism. Daytime observations of an extended source area are also performed; an image motion of small spatial coherent length seems to be drawn by a horizontal wind producing wave-like distortion of horizontal lines and boiling-like of vertical ones.
The object of this communication is to compare two inversion algorithms in their application to the blind deconvolution problem. After a brief summary of the previous works in this field, we describe the Richardson-Lucy and the steepest descent algorithms and we introduce these methods in the basic error reduction algorithm of Ayers and Dainty. These algorithms are compared when used for blind deconvolution of simulated binary objects convolved by a point spread function and corrupted by a Gaussian additive noise. We consider the effects of the noise level on the reconstruction error, together with the effects of the algorithmic parameters (inner and outer iteration numbers). Particular effects occurring during the reconstruction process are also shown.
The aim of this communication is to show how the Richardson-Lucy deconvolution algorithm can be applied to the blind deconvolution problem. After a brief description of the R.L algorithm itself, we start from the basic papers of Ayers and Dainty (1988) and Lane (1992) and introduce in their approach the R.L algorithm in several different ways. We show that the general behaviour of the proposed methods is analogous to that of the error-reduction algorithms and that good solutions can be obtained. The unregularized behaviour of the RL algorithm is overcame by a limitation of the iteration number. Moreover we compare the structures of the various algorithms proposed here and emphasise the main differences. The proposed algorithms are used to blindly deconvolve two types of objects (point-like and extended objects) blurred by simulated point spread functions similar to those observed at the focus of a small telescope in presence of at- mospheric turbulence. The error reduction term is given as a function of the iteration number.