Invariant representations in deep learning for optoacoustic imaging

Image reconstruction in optoacoustic tomography (OAT) is a trending learning task highly dependent on measured physical magnitudes present at sensing time. A large number of different settings and also the presence of uncertainties or partial knowledge of parameters can lead to reconstruction algorithms that are specifically tailored and designed to a particular configuration, which could not be the one that will ultimately be faced in a final practical situation. Being able to learn reconstruction algorithms that are robust to different environments (e.g., the different OAT image reconstruction settings) or invariant to such environments is highly valuable because it allows us to focus on what truly matters for the application at hand and discard what are considered spurious features. In this work, we explore the use of deep learning algorithms based on learning invariant and robust representations for the OAT inverse problem. In particular, we consider the application of the ANDMask scheme due to its easy adaptation to the OAT problem. Numerical experiments are conducted showing that when out-of-distribution generalization (against variations in parameters such as the location of the sensors) is imposed, there is no degradation of the performance and, in some cases, it is even possible to achieve improvements with respect to standard deep learning approaches where invariance robustness is not explicitly considered.

On the robustness of model-based algorithms for photoacoustic tomography: Comparison between time and frequency domains

For photoacoustic image reconstruction, certain parameters such as sensor positions and speed of sound have a major impact on the reconstruction process and must be carefully determined before data acquisition. Uncertainties in these parameters can lead to errors produced by a modeling mismatch, hindering the reconstruction process and severely affecting the resulting image quality. Therefore, in this work, we study how modeling errors arising from uncertainty in sensor locations affect the images obtained by matrix model-based reconstruction algorithms based on time domain and frequency domain models of the photoacoustic problem. The effects on the reconstruction performance with respect to the uncertainty in the knowledge of the sensors location are compared and analyzed both in a qualitative and quantitative fashion for both time and frequency models. Ultimately, our study shows that the frequency domain approach is more sensitive to this kind of modeling errors. These conclusions are supported by numerical experiments and a theoretical sensitivity analysis of the mathematical operator for the direct problem.

Single site alpha-tubulin mutation affects astral microtubules and nuclear positioning during anaphase in Saccharomyces cerevisiae: possible role for palmitoylation of alpha-tubulin

We generated a strain of Saccharomyces cerevisiae in which the sole source of alpha-tubulin protein has a cys-to-ser mutation at cys-377, and then we examined microtubule morphology and nuclear positioning through the cell cycle. During G1 of the cell cycle, microtubules in the C377S alpha-tubulin (C377S tub1) mutant were indistinguishable from those in the control (TUB1) strain. However, mitotic C377S tub1 cells displayed astral microtubules that often appeared excessive in number, abnormally long, and/or misoriented compared with TUB1 cells. Although mitotic spindles were always correctly aligned along the mother-bud axis, translocation of spindles through the bud neck was affected. In late anaphase, spindles were often not laterally centered but instead appeared to rest along the sides of cells. When the doubling time was increased by growing cells at a lower temperature (15 degrees C), we often found abnormally long mitotic spindles. No increase in the number of anucleate or multinucleate C377S mutant cells was found at any temperature, suggesting that, despite the microtubule abnormalities, mitosis proceeded normally. Because cys-377 is a presumptive site of palmitoylation in alpha-tubulin in S. cerevisiae, we next compared in vivo palmitoylation of wild-type and C377S mutant forms of the protein. We detected palmitoylated alpha-tubulin in TUB1 cells, but the cys-377 mutation resulted in approximately a 60% decrease in the level of palmitoylated alpha-tubulin in C377S tub1 cells. Our results suggest that cys-377 of alpha-tubulin, and possibly palmitoylation of this amino acid, plays a role in a subset of astral microtubule functions during nuclear migration in M phase of the cell cycle.

Function of tubulin binding proteins in vivo

Overexpression of the beta-tubulin binding protein Rbl2p/cofactor A is lethal in yeast cells expressing a mutant alpha-tubulin, tub1-724, that produces unstable heterodimer. Here we use RBL2 overexpression to identify mutations in other genes that affect formation or stability of heterodimer. This approach identifies four genes-CIN1, CIN2, CIN4, and PAC2-as affecting heterodimer formation in vivo. The vertebrate homologues of two of these gene products-Cin1p/cofactor D and Pac2p/cofactor E-can catalyze exchange of tubulin polypeptides into preexisting heterodimer in vitro. Previous work suggests that both Cin2p or Cin4p act in concert with Cin1p in yeast, but no role for vertebrate homologues of either has been reported in the in vitro reaction. Results presented here demonstrate that these proteins can promote heterodimer formation in vivo. RBL2 overexpression in cin1 and pac2 mutant cells causes microtubule disassembly and enhanced formation of Rbl2p-beta-tubulin complex, as it does in the alpha-tubulin mutant that produces weakened heterodimer. Significantly, excess Cin1p/cofactor D suppresses the conditional phenotypes of that mutant alpha-tubulin. Although none of the four genes is essential for viability under normal conditions, they become essential under conditions where the levels of dissociated tubulin polypeptides increase. Therefore, these proteins may provide a salvage pathway for dissociated tubulin heterodimers and so rescue cells from the deleterious effects of free beta-tubulin.

An alpha-tubulin mutant destabilizes the heterodimer: phenotypic consequences and interactions with tubulin-binding proteins

Many effectors of microtubule assembly in vitro enhance the polymerization of subunits. However, several Saccharomyces cerevisiae genes that affect cellular microtubule-dependent processes appear to act at other steps in assembly and to affect polymerization only indirectly. Here we use a mutant alpha-tubulin to probe cellular regulation of microtubule assembly. tub1-724 mutant cells arrest at low temperature with no assembled microtubules. The results of several assays reported here demonstrate that the heterodimer formed between Tub1-724p and beta-tubulin is less stable than wild-type heterodimer. The unstable heterodimer explains several conditional phenotypes conferred by the mutation. These include the lethality of tub1-724 haploid cells when the beta-tubulin-binding protein Rbl2p is either overexpressed or absent. It also explains why the TUB1/tub1-724 heterozygotes are cold sensitive for growth and why overexpression of Rbl2p rescues that conditional lethality. Both haploid and heterozygous tub1-724 cells are inviable when another microtubule effector, PAC2, is overexpressed. These effects are explained by the ability of Pac2p to bind alpha-tubulin, a complex we demonstrate directly. The results suggest that tubulin-binding proteins can participate in equilibria between the heterodimer and its components.

Formation and function of the Rbl2p-beta-tubulin complex

The yeast protein Rbl2p suppresses the deleterious effects of excess beta-tubulin as efficiently as does alpha-tubulin. Both in vivo and in vitro, Rbl2p forms a complex with beta-tubulin that does not contain alpha-tubulin, thus defining a second pool of beta-tubulin in the cell. Formation of the complex depends upon the conformation of beta-tubulin. Newly synthesized beta-tubulin can bind to Rbl2p before it binds to alpha-tubulin. Rbl2p can also bind beta-tubulin from the alpha/beta-tubulin heterodimer, apparently by competing with alpha-tubulin. The Rbl2p-beta-tubulin complex has a half-life of approximately 2.5 h and is less stable than the alpha/beta-tubulin heterodimer. The results of our experiments explain both how excess Rbl2p can rescue cells overexpressing beta-tubulin and how it can be deleterious in a wild-type background. They also suggest that the Rbl2p-beta-tubulin complex is part of a cellular mechanism for regulating the levels and dimerization of tubulin chains.

Barriers to health care access for Latino children: a review

BACKGROUND AND OBJECTIVES

More than 9 million Latino children currently live in the United States. Latinos will soon be the largest minority group in the country, but little is known about access barriers to health care faced by Latino children. We reviewed the literature to define specific barriers to care for Latino children, identify methodologic problems, and highlight the clinical and research implications of the identified barriers.

METHODS

We did a MEDLINE search, using combinations of the key words Hispanic, children, and access. Study exclusion criteria included "not an original research article," "enrolled only adult subjects," "no separate data analysis for children," and "dental care focus."

RESULTS

The search yielded 497 citations, of which 27 met the inclusion criteria. Of the 32 potential barriers identified, 21 had good supportive evidence. Lack of health insurance was a consistent barrier; recent data revealed that 26% of Latino children are uninsured, compared with 10% of white children and 14% of African-American children. Latino children also are at greater risk for episodic insurance coverage, low rates of private insurance, and loss of employee-based coverage. Parent beliefs about the etiology and treatment of their child's illness, use of home remedies, choice of sources of advice, and folk medicine practices may also influence how health care is obtained. Few data are available on differences in access among major Latino subpopulations, and no studies focused primarily on barriers as perceived by Latino parents. Evidence is equivocal or lacking that the following are barriers for Latino children: immigration status, duration of parent residency in the United States, and acculturation. Several barriers were identified that originate with practices and behaviors of health care providers, including reduced screening, missed vaccination opportunities, decreased likelihood of receiving prescriptions, and poor communication.

CONCLUSIONS

Lack of health insurance and lack of a regular source of care are major access barriers for Latino children, but many other barriers were identified that also can have a substantial effect on health care. In addition, the behaviors and practices of both health care providers and parents can affect access to care. Too little is known about what parents perceive to be the major barriers, access differences among Latino subpopulations, the roles of language and culture, and the causes of obstacles resulting from the actions of providers.

Rbl2p, a yeast protein that binds to beta-tubulin and participates in microtubule function in vivo

Genetic configurations resulting in high ratios of beta-tubulin to alpha-tubulin are toxic in S. cerevisiae, causing microtubule disassembly and cell death. We identified three non-tubulin yeast genes that, when overexpressed, rescue cells from excess beta-tubulin. One, RBL2, rescues beta-tubulin lethality as efficiently as does alpha-tubulin. Rbl2p binds to beta-tubulin in vivo. Deficiencies or excesses of either Rbl2p or alpha-tubulin affect microtubule-dependent functions in a parallel fashion. Rbl2p has functional homology with murine cofactor A, a protein important for in vitro assays of beta-tubulin folding. The results suggest that Rbl2p participates in microtubule morphogenesis but not in the assembled polymer.

A family of U1 pseudogenes in Bombyx mori may be derived from an ancestral pseudogene

Seven EMBL-4 lambda clones containing U1 small nuclear RNA sequences were isolated from a Bombyx mori genomic library. Six of the seven represent unique sequences. The six unique U1 sequences exhibit fixed point 3'-end truncation. Five out of the six clones share immediate 3'-end flanking sequences while two share 5'-end flanking sequences. Fixed point 3'-end truncation and a hierarchy of shared to unique diagnostic mutations may suggest a family of U1 pseudogenes generated from a reverse-transcribed class II pseudogene in B. mori. An ancestral 'master' U1 pseudogene capable of RNA- and/or DNA-mediated transposition may give rise to generations of U1 pseudogenes that include the original pseudogene's flanking sequences. Identical 3'-end truncation in some of these U1 sequences can be explained by RNA self-priming due to intra-strand binding prior to reverse transcription.

Enrichment of biologically active U1 small nuclear RNAs by ion-exchange high-performance liquid chromatography

The use of ion-exchange high-performance liquid chromatography in conjunction with preparative electrophoresis to facilitate the purification of biologically active snRNAs is described. Separation of total nuclear RNA from a Bombyx mori cell line was done with a Bio-Rad MA7 plasmid column in a HRLC 500 system. Individual fractions were subjected to electrophoresis through 14% polyacrylamide gels for identification. High levels of U1 RNA were confirmed by Northern analysis with a human U1 probe. Biological activity of RNAs from the column was demonstrated by their ability to incorporate 32P-AMP at the 3' end. Ion-exchange chromatography provides a rapid, automated method for purifying large amounts of RNAs that can then be utilized in further studies.