A 3D Probabilistic Deep Learning System for Detection and Diagnosis of Lung Cancer Using Low-Dose CT Scans.

We introduce a new computer aided detection and diagnosis system for lung cancer screening with low-dose CT scans that produces meaningful probability assessments. Our system is based entirely on 3D convolutional neural networks and achieves state-of-the-art performance for both lung nodule detection and malignancy classification tasks on the publicly available LUNA16 and Kaggle Data Science Bowl challenges. While nodule detection systems are typically designed and optimized on their own, we find that it is important to consider the coupling between detection and diagnosis components. Exploiting this coupling allows us to develop an end-to-end system that has higher and more robust performance and eliminates the need for a nodule detection false positive reduction stage. Furthermore, we characterize model uncertainty in our deep learning systems, a first for lung CT analysis, and show that we can use this to provide well-calibrated classification probabilities for both nodule detection and patient malignancy diagnosis. These calibrated probabilities informed by model uncertainty can be used for subsequent risk-based decision making towards diagnostic interventions or disease treatments, as we demonstrate using a probability-based patient referral strategy to further improve our results.

Composite organic-inorganic nanoparticles as Raman labels for tissue analysis.

Composite organic-inorganic nanoparticles (COINs) are novel optical labels for detection of biomolecules. We have previously developed methods to encapsulate COINs and to functionalize them with antibodies. Here we report the first steps toward application of COINs to the detection of proteins in human tissues. Two analytes, PSA and CK18, are detected simultaneously using two different COINs in a direct binding assay, and two different COINs are shown to simultaneously label PSA in tissue samples.

Ultrasensitive detection and characterization of posttranslational modifications using surface-enhanced Raman spectroscopy.

Posttranslational modification (PTM) of proteins is likely to be the most common mechanism of altering the expression of genetic information. It is essential to characterize PTMs to establish a complete understanding of the activities of proteins. Here, we present a sensitive detection method using surface-enhanced Raman spectroscopy (SERS) that can detect PTMs from as little as zeptomoles of peptide. We demonstrate, using model peptides, the ability of SERS to detect a variety of protein modifications, such as acetylation, trimethylation, phosphorylation, and ubiquitination. In addition, we show the capability to obtain positional information for modifications such as trimethylation and phosphorylation using SERS and wavelet decomposition data analysis techniques. We further show that it is possible to apply SERS to detect PTMs from biological samples such as histones. We envision that this detection method might be a valuable technique that is complementary to mass spectrometry in obtaining orthogonal chemical and modification-specific information from biological samples at sensitive levels.

Single-molecule detection of biomolecules by surface-enhanced coherent anti-Stokes Raman scattering.

We report on the applicability of combining surface-enhanced Raman scattering (SERS) with coherent anti-Stokes Raman scattering for high-sensitivity detection of biological molecules. We found that this combination of techniques provides more than 3 orders of signal enhancement compared with SERS and permits monitoring of biological molecules such as deoxyguanosine monophosphate (dGMP) and deoxyadenosine monophosphate at the single-molecule level. This combined technique also improved detection sensitivity for angiotensin peptide. As this is believed to be the first report of detection of dGMP at the single-molecule level, we suggest that this approach can serve as a new tool for biological studies.

Composite organic-inorganic nanoparticles (COINs) with chemically encoded optical signatures.

To obtain a coding system for multiplex detection, we have developed a method to synthesize a new type of nanomaterial called composite organic-inorganic nanoparticles (COINs). The method allows the incorporation of a broad range of organic compounds into COINs to produce surface enhanced Raman scattering (SERS)-like spectra that are richer in variety than fluorescence-based signatures. Preliminary data suggest that COINs can be used as Raman tags for multiplex and ultrasensitive detection of biomolecules.

Microfluidic operations using deformable polymer membranes fabricated by single layer soft lithography.

We show that it is possible to use single layer soft lithography to create deformable polymer membranes within microfluidic chips for performing a variety of microfluidic operations. Single layer microfluidic chips were designed, fabricated, and characterized to demonstrate pumping, sorting, and mixing. Flow rates as high as 0.39 microl min(-1) were obtained by peristaltic pumping using pneumatically-actuated membrane devices. Sorting was attained via pneumatic actuation of membrane units placed alongside the branch channels. An active mixer was also demonstrated using single-layer deformable membrane units.

Specific chemical effects on surface-enhanced Raman spectroscopy for ultra-sensitive detection of biological molecules.

Achieving high signal amplification in surface-enhanced Raman scattering (SERS) is important for reaching single molecule level sensitivity and has been the focus of intense research efforts. We introduce a novel chemical enhancer, lithium chloride, that provides an additional order of magnitude increase in SERS relative to previously reported enhancement results. We have duplicated single molecule detection of the DNA base adenine that has previously been reported, thereby providing independent validation of this important result. Building upon this work, we show that the chemical enhancer LiCl produces strong SERS signal under a wide range of experimental conditions, including multiple laser excitation wavelengths and target molecule concentrations, for nucleotides, nucleosides, bases, and dye molecules. This is significant because while selection of anions used in chemical enhancement is well known to affect the degree of amplification attained, cation selection has previously been reported to have no major effect on the magnitude of SERS enhancement. Our findings indicate that cation selection is quite important in ultra-sensitive SERS detection, opening the door to further discussion and theory development involving the role of cations in SERS.