Biosketch for Lawrence Lifshitz
Education
Institution | Degree | Date | Field of Study |
Harvard University | B.A. | 1980 | Physics |
Univ. of N.C. at Chapel Hill | M.S. | 1983 | Computer Science |
Univ. of N.C. at Chapel Hill | Ph.D. | 1987 | Computer Science |
Personal Statement
I have been a member of the Biomedical Imaging Group at UMMS for the last 29 years. I have expertise in the areas of computer graphics, computer vision, simulation, and statistics. I am the principal developer of a number of visualization and analysis software systems and hundreds of programs. The Data Analysis and Visualization Environment (DAVE) program I developed permits interactive 3D volume and surface visualization (in addition to many other capabilities, including ones for remote collaboration and quantification). Other programs deal with (among other things) reaction-diffusion and channel simulation models, statistics for spatial randomness of channels, deformable models, colocalization analysis, and tracking intracellular vesicles to quantify exocytosis. Additional information about some of the software I've written is available online. A much more complete list of BIG software is also available .
Positions and Employment
- 1980-1987 Research Asst. In Computer Graphics & Computer Vision, Computer Science Dept. UNC, Chapel Hill, NC.
- 1987-2002 Assistant Professor, Biomedical Imaging Group, Depts. of Physiology & Nuclear Medicine, Univ. of Mass. Medical School,Worcester, MA
- 2002-2008 Associate Professor, Biomedical Imaging Group, Dept. of Cellular & Molecular Physiology, Univ. of Mass. Medical School, Worcester, MA
- 2008-present Associate Professor, Biomedical Imaging Group, Program in Molecular Medicine, Univ. of Mass. Medical School, Worcester, MA
Contribution to Science
I have used my broad expertise across several technical fields, combined with my strong collaborative skills, to design new algorithms and software to help answer biological questions posed by my many collaborators. This has resulted in over 45 refereed journal publications and sections in 5 books. There are a few broad strands which can be discerned in the research I've performed over the years.
-
Computer vision
Computer vision is the finding of objects in images, usually with the goal of then quantifying some aspect of the object itself (e.g. Its size or position). My dissertation [Lifshitz] involved automated analysis of CT images to find organs using scale-space analysis – a field which has since blossomed.
I was also involved in some early research in deformable models – developing, with collaborators, a simple, fast, deformable model with which to find the plasma membrane of smooth muscle cells. This model was unusual in that it permitted local constraints to be easily applied to it, making it suitable for many tasks. Finding the plasma membrane in a cell using a deformable model permitted us to restrict analysis to molecules (e.g. channels) on or near the membrane on several research projects; one such project was [Moore]. It used the deformable model to eliminate nonspecific label in the interior of the cell and was therefore able to perform a precise calculation of colocalization of the Na+/K+ pump, Na+/Ca2+ exchanger, Vinculin (a marker for contractile filaments), and calsequestrin (marking Ca2+ in SR stores).
- a. Lifshitz LM, Pizer SM, “A Multiresolution Hierarchical Approach to Image Segmentation Based on Intensity Extrema”, IEEE Transactions on Pattern Analysis and Machine Intelligence. Vol. 12, No. 6, June, 1990, pp. 529-540. Cited > 300 times.
- b. Moore E, Etter E, Philipson K, Carrington W, Fogarty K, Lifshitz LM, Fay FS, “Coupling of the Na+/Ca+2 Exchanger to the Na+/K+ Pump and Sarcoplasmic Reticulum in Smooth Muscle”, Nature, Vol. 365, October 14, 1993, pp. 657-660. PMID: 8413629.
-
Tracking
In live cells, after being found an object (e.g., microtubule, exocytic vesicle) often needs to be tracked over time. In [Lifshitz] I used an interpretation tree with an unusual set of cost functions to find and follow several microtubules in a 2D time series of a moving cell. An interpretation tree was used to match fragments of microtubules (identified using a type of edge detector) to a model of the microtubules obtained manually from the first image in the time series and then propagated forward in time. Costs of a match included the cost of not matching (i.e., leaving a model microtubule unmatched incurred a high cost even if all fragments matched their microtubules well). This unusual cost structure was critical.
Total internal reflection fluorescence (TIRF) microscopy revealed highly mobile vesicles containing enhanced green fluorescent protein-tagged glucose transporter (GLUT4) within a zone about 100 nm beneath the plasma membrane of 3T3-L1 adipocytes. In [Huang] we developed a computer program that enabled direct analysis of the docking/fusion kinetics of hundreds of exocytic-vesicles fusing to the plasma membrane. This required automated detection of the vesicles, tracking them through the fusion event and fitting kinetics to the observed events. We determined that insulin stimulation increases the fusion frequency of exocytic GLUT4 vesicles by approximately 4-fold.
- a. Lifshitz LM, “Model-Based Tracking of Deformable Filaments”, SPIE Proceedings, vol. 1609, Model-Based Vision Development and Tools, Larson RM and Nasr HN (eds.), November, 1991, pp. 185-197
- b. Huang S, Lifshitz LM, Jones C, Bellve KD, Standley C, Fonseca S, Corvera S, Fogarty KE, Czech MP. “Insulin stimulates membrane fusion and GLUT4 accumulation in clathrin coats on adipocyte plasma membranes”. Mol Cell Biol. 2007 May;27(9):3456-69. Epub 2007 Mar 5. PubMed PMID: 17339344; PubMed Central PMCID: PMC1899973
-
Computer Graphics
I am the principal developer of the Data Analysis and Visualization Environment ( DAVE ) program [Lifshitz]. DAVE permits interactive 3D volume visualization and quantification of microscopy data (in 3 colors over time). All of the features available in DAVE are too numerous to detail, arising from more than 100,000 lines of C code. DAVE was unusual in representing volume data (voxels) as tiny polygons. This permitted both surface data (e.g., a deformable surface model) and volume data (e.g., a 3D microscope image of a cell) to be displayed in a unified manner with both taking advantage of the fast hardware for rendering polygons just becoming available on SGI computers at that time. DAVE has been used by the Biomedical Imaging Group on a daily basis for long periods of time to aid many research projects. For instance, in [Rizzuto] DAVE was used to observe the intertwining of mitochondria and ER. I was then able to determine that 5-20% of the mitochondrial surface was close to the ER (a calcium store). This was a seminal paper exposing the mitochondria's role in calcium handling. DAVE has produced images for many of the articles published by the Biomedical Imaging Group and others; including the cover images of many journals. The [Femino] cover image in Science depicted one of the first visualizations of a single RNA inside a cell.
- a. Lifshitz LM, Collins J, Moore E, Gauch J, "Computer Vision and Graphics in Fluorescence Microscopy". In: Proceedings of the IEEE Workshop on Biomedical Image Analysis, IEEE Computer Society Press, Los Alamitos, CA, 1994, pp. 166-175
- b. Rizzuto R., Pinton P, Carrington W, Fay F, Fogarty K, Lifshitz L, Tuft R, Pozzan T, ”Close Contacts with the Endoplasmic Reticulum as Determinants of Mitochondrial Ca2+ Responses”, Science, Vol. 280, June 12, 1998, pp. 1763-1766. PubMed PMID: 9624056. Cited >1,300 times.
- c. Femino AM, Fay FS, Fogarty KE, Singer RH. “Visualization of Single RNA Transcripts in Situ”. Science vol. 280, pp. 585-590. 1998. April. Cited > 500 times. Cover photo from DAVE.
- d. Via a web browser you can see both the DAVE manual and some DAVE images .
-
Statistics/Colocalization
Much of the quantitation I've performed requires the creation of a statistical test, often using Monte Carlo techniques, to answer biological questions in a rigorous manner. In [Fay] the spatial relationship between a newly synthesized RNA (BrUTP) and an RNA splicing factor (SC35) was examined. We determined that significant true signal (~70%) of the SC35 was represented as low level diffuse signal rather than the (2x) brighter "speckles" typically analyzed. Percent colocalization of BrUTP with SC35 was calculated. The statistical significance of this colocalization was calculated by ranking it relative to colocalizations obtained via random translations of the two data sets (importantly preserving any spatial correlations within each image). This approach highlighted both the importance of dim diffuse signal which was often ignored by researchers and that low levels of colocalization could still be much higher than that expected due to random chance.
[Lifshitz] describes methods developed to study the spatial distribution of Protein Kinase C and Vinculin along the plasma membrane. A deformable model was used to restrict analysis to the plasma membrane. I extended the standard K function test for complete spatial randomness (CSR) to apply to voxelized membranes in 3D (standard theory applies to points on a flat plane). In addition, I developed a method to determine whether the two distributions were independent of each other; this extended flat 2D Monte Carlo methods to random translations of the data sets along the plasma membrane. Results showed that PKC and Vinculin were individually non-randomly distributed and also not randomly distributed relative to each other. In [Bassell] I developed a statistical test to determine if B-actin mRNA preferentially located near microtubules. The mRNA was point-like and the micotubules line-like. The Monte-Carlo test I designed compared the histogram of observed distances from mRNA to microtubule with histograms from randomly placed points.
- a. Fay FS, Taneja KL, Shenoy S, Lifshitz LM, Singer RH, “Quantitative digital analysis of diffuse and concentrated distributions of nascent transcripts, SC35 and Poly(A)”, Experimental Cell Research (with cover photo), Vol. 231(1), February 25, 1997, pp. 27-37. PMID: 9056409. Cited > 100 times.
- b. Lifshitz LM, “Determining data independence on a digitized membrane in three dimensions”, IEEE transactions on medical imaging. 1998; 17(2):299-303. PMID: 9688162
- c. Bassell G, Zhang H, Byrd A, Femino A , Singer R , Taneja K, Lifshitz L, Herman I, Kosik K, ”Sorting of beta-Actin mRNA and Protein to Neurites and Growth Cones in Culture”, Journal of Neuroscience, Vol. 18(1), January, 1998, pp.251-265. PMID: 9412505. Cited > 350 times.
-
Mathematical Modeling/Simulation
I have developed mathematical software to model and simulate various aspects of intracellular function. Discrete localized fluorescence transients due to openings of a single plasma membrane Ca2+ permeable cation channel were recorded in [Zou]. The equations which model this Ca2+ influx and binding are reaction-diffusion equations. I wrote a program to simulate this in a cylindrical smooth muscle cell. Using cylindrical symmetry reduced simulation time from days to hours. This and related software has been used to model other Ca2+ release events in the cell. In [ZhuGe] I used it to study Ca2+ sparks; these are highly localized, transient releases of Ca2+ from sarcoplasmic reticulum through ryanodine receptors. In [Bao] we investigated the spatial organization of ryanodine receptors and Chloride channels in airway myocytes from mouse. Meanwhile, [Lifshitz] used the spatial relationship (gotten via 3D imaging) between ryanodine receptors (the Ca2+ source for sparks) and BK channels (the Ca2+ target), the spatial-temporal profile of [Ca2+] resulting from Ca2+ sparks, and a kinetic model of BK channels, all incorporated into my simulation software, to estimate that an average Ca2+ spark caused by the opening of a cluster of ryanodine receptors acts on BK channels situated in two to three clusters that are randomly distributed within an ~600-nm radius of rynodine receptors.
- a. Zou H, Lifshitz L, Tuft R, Fogarty K, Singer J, “Imaging Ca2+ Entering the Cytoplasm through a Single Opening of a Plasma Membrane Cation Channel”, Journal of General Physiology, vol. 114, Oct. 1999, pp. 1-14. PMID: 10498675; PMCID: PMC2229469. Cover photo produced with DAVE.
- b. ZhuGe R, Fogarty KE, Tuft RA, Lifshitz LM, Sayar K, Walsh JV Jr. “Dynamics of signaling between Ca(2+) sparks and Ca(2+) activated K(+) channels studied with a novel image-based method for direct intracellular measurement of ryanodine receptor Ca(2+) current”, J Gen Physiol. 2000 Dec;116(6):845-64. PMID: 11099351; PMCID: PMC2231814.
- c. Bao R, Lifshitz LM, Tuft RA, Bellvé K, Fogarty KE, ZhuGe R. “A close association of RyRs with highly dense clusters of Ca2+-activated Cl- channels underlies the activation of STICs by Ca2+ sparks in mouse airway smooth muscle". J Gen Physiol. 2008 Jul;132(1):145-60. doi: 10.1085/jgp.200709933. PMID: 18591421; PMCID: PMC2442178
- d. Lifshitz LM, Carmichael JD, Lai FA, Sorrentino V, Bellvé K, Fogarty KE, ZhuGe R. “Spatial organization of RYRs and BK channels underlying the activation of STOCs by Ca(2+) sparks in airway myocytes". J Gen Physiol. 2011 Aug;138(2):195-209. doi: 10.1085/jgp.201110626. Epub 2011 Jul 11. PMID: 21746845; PMCID: PMC3149436.
Complete List of Published Work
NCBI Bibliography