AI and human society

This paper considers the impact of the AI R&D programme on human society and the individual human being on the assumption that a full realisation of the engineering objective of AI, namely, construction of human-level, domain-independent intelligent entities, is possible. Our assumption is essentially identical tothe maximum progress scenario of the Office of Technology Assessment, US Congress.Specifically, the first section introduces some of the significant issues on the relational nexus among work, education and the human-machine boundary. In particular, based on a Russellian conception of rationality I briefly argue that we need to change our related conceptions of work, employment and free time, through a new human-centred education. On the human-machine boundary problem, I make a couple of tentative suggestions and put forward some crucial open questions.Section two discusses the impact of the emerging machine intelligence on human nature both as modification of its self-image, keeping human nature itself unchanged, and its potential for altering human nature itself. I briefly argue that: (i) in a certain context, the question of the supremacy or uniqueness of human intelligence loses much, if not all, of its weight; and (ii) appearance of Robot-X species would immortalise the human spirit.     

Effect of plasma TNF-alpha on filgrastim-stimulated hematopoiesis in mice and humans

Marshall syndrome is a rare, autosomal dominant skeletal dysplasia that is phenotypically similar to the more common disorder Stickler syndrome. For a large kindred with Marshall syndrome, we demonstrate a splice-donor-site mutation in the COL11A1 gene that cosegregates with the phenotype. The G+1-->A transition causes in-frame skipping of a 54-bp exon and deletes amino acids 726-743 from the major triple-helical domain of the alpha1(XI) collagen polypeptide. The data support the hypothesis that the alpha1(XI) collagen polypeptide has an important role in skeletal morphogenesis that extends beyond its contribution to structural integrity of the cartilage extracellular matrix. Our results also demonstrate allelism of Marshall syndrome with the subset of Stickler syndrome families associated with COL11A1 mutations.     

Multigrid geometric active contour models.

Geometric active contour models are very popular partial differential equation-based tools in image analysis and computer vision. We present a new multigrid algorithm for the fast evolution of level-set-based geometric active contours and compare it with other established numerical schemes. We overcome the main bottleneck associated with most numerical implementations of geometric active contours, namely the need for very small time steps to avoid instability, by employing a very stable fully 2-D implicit-explicit time integration numerical scheme. The proposed scheme is more accurate and has improved rotational invariance properties compared with alternative split schemes, particularly when big time steps are utilized. We then apply properly designed multigrid methods to efficiently solve the occurring sparse linear system. The combined algorithm allows for the rapid evolution of the contour and convergence to its final configuration after very few iterations. Image segmentation experiments demonstrate the efficiency and accuracy of the method.     

On l∞ and l2 robustness of spatially invariant systems

We consider spatiotemporal systems and study their l_∞ and l₂ robustness properties in the presence of spatiotemporal perturbations. In particular, we consider spatially invariant nominal models and provide necessary and sufficient conditions for system robustness for the cases when the underlying perturbations are linear spatiotemporal varying, and nonlinear spatiotemporal invariant, unstructured or structured. It turns out that these conditions are analogous to the scaled small gain condition (which is equivalent to a spectral radius condition and a linear matrix inequality for the l_∞ and l₂ cases, respectively) derived for standard linear time‐invariant models subject to time‐varying linear and time‐invariant nonlinear perturbations. Copyright © 2009 John Wiley & Sons, Ltd.     

Temporal course of perfusion in human masseter muscle during isometric contraction assessed by arterial spin labeling at 3T

Object  

Examination of blood perfusion in the masseter muscle in the course of repetitive isometric contraction by arterial spin-labeling (ASL) MR imaging and additional T2 relaxation time measurements during and after masseter muscle activation.     

Automated reference-free detection of motion artifacts in magnetic resonance images

Objectives

Our objectives were to provide an automated method for spatially resolved detection and quantification of motion artifacts in MR images of the head and abdomen as well as a quality control of the trained architecture.

Materials and methods

T1-weighted MR images of the head and the upper abdomen were acquired in 16 healthy volunteers under rest and under motion. Images were divided into overlapping patches of different sizes achieving spatial separation. Using these patches as input data, a convolutional neural network (CNN) was trained to derive probability maps for the presence of motion artifacts. A deep visualization offers a human-interpretable quality control of the trained CNN. Results were visually assessed on probability maps and as classification accuracy on a per-patch, per-slice and per-volunteer basis.

Results

On visual assessment, a clear difference of probability maps was observed between data sets with and without motion. The overall accuracy of motion detection on a per-patch/per-volunteer basis reached 97%/100% in the head and 75%/100% in the abdomen, respectively.

Conclusion

Automated detection of motion artifacts in MRI is feasible with good accuracy in the head and abdomen. The proposed method provides quantification and localization of artifacts as well as a visualization of the learned content. It may be extended to other anatomic areas and used for quality assurance of MR images.

     

A comparison of emulsifiers for the formation of oil-in-water emulsions: stability of the emulsions within 9 h after production and MR signal properties

Objective

To provide a basis for the selection of suitable emulsifiers in oil-in-water emulsions used as tissue analogs for MRI experiments. Three different emulsifiers were investigated with regard to their ability to stabilize tissue-like oil-in-water emulsions. Furthermore, MR signal properties of the emulsifiers themselves and influences on relaxation times and ADC values of the aqueous phase were investigated.

Materials and methods

Polysorbate 60, sodium dodecyl sulfate (SDS) and soy lecithin were used as emulsifiers. MR characteristics of emulsifiers were assessed in aqueous solutions and their function as a stabilizer was examined in oil-in-water emulsions of varying fat content (10, 20, 30, 40, 50%). Stability and homogeneity of the oil-in-water emulsions were evaluated with a delay of 3 h and 9 h after preparation using T₁ mapping and visual control. Signal properties of the emulsifiers were investigated by ¹H-MRS in aqueous emulsifier solutions. Relaxometry and diffusion weighted MRI (DWI) were performed to investigate the effect of various emulsifier concentrations on relaxation times (T₁ and T₂) and ADC values of aqueous solutions.

Results

Emulsions stabilized by polysorbate 60 or soy lecithin were stable and homogeneous across all tested fat fractions. In contrast, emulsions with SDS showed a significantly lower stability and homogeneity. Recorded T₁ maps revealed marked creaming of oil droplets in almost all of the emulsions with SDS. The spectral analysis showed several additional signals for polysorbate and SDS. However, lecithin remained invisible in ¹H-MRS. Relaxometry and DWI revealed different influences of the emulsifiers on water: Polysorbate and SDS showed only minor effects on relaxation times and ADC values of aqueous solutions, whereas lecithin showed a strong decrease in both relaxation times (r_1,lecithin = 0.11 wt.%⁻¹ s⁻¹, r_2,lecithin = 0.57 wt.%⁻¹ s⁻¹) and ADC value (Δ(ADC)_lecithin =  − 0.18 × 10^–3 mm²/s⋅wt.%) with increasing concentration.

Conclusion

Lecithin is suggested as the preferred emulsifier of oil-in-water emulsions in MRI as it shows a high stabilizing ability and remains invisible in MRI experiments. In addition, lecithin is suitable as an alternative means of adjusting relaxation times and ADC values of water.

     

Diffusion-weighted echo planar MR imaging of the neck at 3 T using integrated shimming: comparison of MR sequence techniques for reducing artifacts caused by magnetic-field inhomogeneities

Objective

Our objective was to compare available techniques reducing artifacts in echo planar imaging (EPI)-based diffusion-weighed magnetic resonance imaging MRI (DWI) of the neck at 3 Tesla caused by B0-field inhomogeneities.

Materials and methods

A cylindrical fat–water phantom was equipped with a Maxwell coil allowing for additional linear B0-field variations in z-direction. The effect of increasing strength of this superimposed gradient on image quality was observed using a standard single-shot EPI-based DWI sequence (sEPI), a zoomed single-shot EPI sequence (zEPI), a readout-segmented EPI sequence (rsEPI), and an sEPI sequence with integrated dynamic shimming (intEPI) on a 3-Tesla system. Additionally, ten volunteers were examined over the neck region using these techniques. Image quality was assessed by two radiologists. Scan durations were recorded.

Results

With increasing strength of the external gradient, marked distortions, signal loss, and failure of fat suppression were observed using sEPI, zEPI, and rsEPI. These artifacts were markedly reduced using intEPI. Significantly better in vivo image quality was also observed using intEPI compared with the other techniques. Scan time of intEPI was similar to sEPI and zEPI and shorter than rsEPI.

Conclusion

The use of integrated 2D shim and frequency adjustment for EPI-based DWI results in a significant improvement in image quality of the head/neck region at 3 Tesla. Combining integrated shimming with rsEPI or zEPI can be expected to provide additional improvements.

     

Comparison of T1-weighted 2D TSE, 3D SPGR,and two-point 3D Dixon MRI for automated segmentation of visceral adipose tissue at 3 Tesla

Objectives

To evaluate and compare conventional T1-weighted 2D turbo spin echo (TSE), T1-weighted 3D volumetric interpolated breath-hold examination (VIBE), and two-point 3D Dixon-VIBE sequences for automatic segmentation of visceral adipose tissue (VAT) volume at 3 Tesla by measuring and compensating for errors arising from intensity nonuniformity (INU) and partial volume effects (PVE).

Materials and methods

The body trunks of 28 volunteers with body mass index values ranging from 18 to 41.2 kg/m² (30.02 ± 6.63 kg/m²) were scanned at 3 Tesla using three imaging techniques. Automatic methods were applied to reduce INU and PVE and to segment VAT. The automatically segmented VAT volumes obtained from all acquisitions were then statistically and objectively evaluated against the manually segmented (reference) VAT volumes.

Results

Comparing the reference volumes with the VAT volumes automatically segmented over the uncorrected images showed that INU led to an average relative volume difference of ?59.22 ± 11.59, 2.21 ± 47.04, and ?43.05 ± 5.01 % for the TSE, VIBE, and Dixon images, respectively, while PVE led to average differences of ?34.85 ± 19.85, ?15.13 ± 11.04, and ?33.79 ± 20.38 %. After signal correction, differences of ?2.72 ± 6.60, 34.02 ± 36.99, and ?2.23 ± 7.58 % were obtained between the reference and the automatically segmented volumes. A paired-sample two-tailed t test revealed no significant difference between the reference and automatically segmented VAT volumes of the corrected TSE (p = 0.614) and Dixon (p = 0.969) images, but showed a significant VAT overestimation using the corrected VIBE images.

Conclusion

Under similar imaging conditions and spatial resolution, automatically segmented VAT volumes obtained from the corrected TSE and Dixon images agreed with each other and with the reference volumes. These results demonstrate the efficacy of the signal correction methods and the similar accuracy of TSE and Dixon imaging for automatic volumetry of VAT at 3 Tesla.