ISBIq: A Framework for Simulation of Cell Cycle in Fluorecence Microscopy. PV182 CBIA seminar

Transcription

1 ISBIq: A Framework for Simulation of Cell Cycle in Fluorecence Microscopy Vladimír Ulman & David Svoboda Centre for Biomedical Image Analysis, Masaryk University Brno, Czech Republic PV182 CBIA seminar March 14, 2013 Ulman & Svoboda (CBIA) ISBIq March 14, / 18

2 Motivation Historically, gtgen and Static simulator were, quite independently, first published/born in Recently, joint proper investigation of this field has begun: Generation of Synthetic Image Datasets for Time-Lapse Fluorescence Microscopy. David Svoboda, and Vladimír Ulman. ICIAR 2, volume 7325 of Lecture Notes in Computer Science, page Springer (2012) We faced several requests for time-lapse datasets during conference meetings and discussions when presenting CytoPacq (static simulator). ISBI Cell Tracking Challenge 2013 Ulman & Svoboda (CBIA) ISBIq March 14, / 18

3 Task a related phase called G 0 ) doing their job in the organism a nerve cell carries impulses, for example. Mitosis is conventionally broken down into five stages: The prophase, aim was toprometaphase, simulate how themetaphase, cell looks during anaphase, the cell andcycle: (M) G 1 MITOTIC PHASE Cytokinesis Mitosis INTERPHASE S (DNA synthesis) Figure 12.6 The cell cycle. In a dividing cell, the mitotic (M) phase alternates with interphase, a growth period. The first part of G 2 near th propha tubule two ce site en tubule the cen Each some h with s tromer site di microt kineto tached crotub When microt pole fr movem posite next is Ulman & Svoboda (CBIA) ISBIq March 14, / 18

4 Problem Analysis Cell cycle can be split into Interphase and Mitosis: Interphase (95%)... simple G1 phase (50%) S phase (30%) G2 phase (15%) Mitosis (5%)... tricky Prophase Metaphase Anaphase Telophase Cytokinesis Note: 100% 24 hour cell cycle length Ulman & Svoboda (CBIA) ISBIq March 14, / 18

5 Problem Analysis Cell cycle phases step by step G2 phase G 2 of Interphase Prophase Prom Centrosomes (with centriole pairs) Nucleolus Nuclear envelope Chromosomes (duplicated, uncondensed) Plasma membrane G 2 of Interphase A nuclear envelope encloses the nucleus. Early mitotic spindle Aster Chromosome, consisting of two sister chromatids Centromere Prophase The chromatin fibers become more Fragments of nuclear envelope cell gather nutrients/energy for mitosis nuclear envelope still encloses nucleus chromosomes have not yet been condensed creation of centrosomes start of mitosis Kinetochore Prom The nuclear enve Ulman & Svoboda (CBIA) ISBIq March 14, / 18

6 Problem Analysis Cell cycle phases step by step Prophase Prophase Prometaphase Early mitotic spindle Aster Chromosome, consisting of two sister chromatids Centromere Fragments of nuclear envelope Kinetochore Nonkinetochore microtubules nucleoli disappear mitotic spindle begins to form (itself) centrosomes move away to opposite poles of cell chromatin condensates each chromosome is composed of two sister chromatids Kinetochore microtubule s. Prophase Prometaphase Ulman The chromatin & Svoboda fibers (CBIA) become more The ISBIq nuclear envelope fragments. March 14, / 18

7 Problem Analysis Cell cycle phases step by step Metaphase Prometaphase Fragments of nuclear envelope Nonkinetochore microtubules nuclear envelope fragments each chromatid is attached to one of two centrosomes chromosomes become more condensed Kinetochore Kinetochore microtubule Prometaphase The nuclear envelope fragments. Ulman & Svoboda (CBIA) ISBIq March 14, / 18

8 Problem Analysis Cell cycle phases step by step Metaphase Metaphase Anaphase Telophase Metaphase plate Cleavage furrow centrocomes are now at opposite poles of the cell chromosomes converge to metaphase plate Spindle Centrosome at one spindle pole Daughter chromosomes Nuclear envelope forming Metaphase Anaphase Te The Ulman centrosomes & Svoboda are (CBIA) now at opposite Anaphase ISBIq is the shortest stage of mitosis, March 14, 2013 Two daughter 5 / 18 n

9 Problem Analysis Cell cycle phases step by step 10 μm Anaphase Anaphase Telophase and Cytokinesis Daughter chromosomes Cleavage furrow Nuclear envelope forming Nucleolus forming (the shortest stage in the cell cycle) two liberated daughter chromosomes begin moving toward opposite ends of cell cell elongates by the end of anaphase, the two ends of the cell have equivalent and complete collection of chromosomes Anaphase Telophase Ulman & Svoboda (CBIA) ISBIq March 14, / 18

10 Problem Analysis Cell cycle phases step by step 10 μm Telophase and Cytokinesis Telophase and Cytokinesis Cleavage furrow Nucleolus forming Nuclear envelope forming two daughter nuclei form in the cell nuclear envelopes arise nucleoli reappear chromosomes become less condensed formation of a cleavage furrow, which pinches the cell in two tosis, Telophase Two daughter nuclei form in the cell. Ulman & Svoboda (CBIA) ISBIq March 14, / 18

11 onsin ize at d the idney Problem Analysis Cell cycle phases step by step Telophase and Cytokinesis Figure Cytokinesis in animal and plant cells. (a) Cleavage of an animal cell (SEM) two daughter nuclei form in the cell ween e the ntact es in segse on Cleavage furrow Contractile ring of microfilaments 100 μm Daughter cells nuclear envelopes arise nucleoli reappear chromosomes become less condensed formation of a cleavage furrow, which pinches the cell in two (b) Cell plate formation in a plant cell (TEM) Ulman & Svoboda (CBIA) ISBIq March 14, / 18

12 Problem Analysis Cell cycle phases step by step G1 phase cell grows to the regular size of cell slowly nucleus still contains only half of all chromosomes chromatin soon decondensates cell gather nutrients/energy for the next (S) phase this is the longest stage in the cell cycle Ulman & Svoboda (CBIA) ISBIq March 14, / 18

13 Problem Analysis Cell cycle phases step by step G1 phase cell grows to the regular size of cell slowly nucleus still contains only half of all chromosomes chromatin soon decondensates cell gather nutrients/energy for the next (S) phase this is the longest stage in the cell cycle Ulman & Svoboda (CBIA) ISBIq March 14, / 18

14 Problem Analysis Cell cycle phases step by step S phase DNA replicates cell keeps to its volume this is the second longest stage in the cell cycle Ulman & Svoboda (CBIA) ISBIq March 14, / 18

15 Design of the new simulator Model of cell The following data structures were used to model the cell phantom cell... binary mask nucleus... binary mask nucleoli... binary mask chromosomes... sets of dots (molecules of fluorescent dye) Note: Aside to binary masks, lists of boundary points were used as well. The following (basic) conditions are valid number of chromosomes (46) during interphase, the chromosomes are located inside nucleus during mitosis, the nucleus disappears and chromosomes are spread over the whole cell Ulman & Svoboda (CBIA) ISBIq March 14, / 18

16 Design of the new simulator Single cell simulation The Cell::DoNextPhase() template Simulates particular cell cycle phase completely and atomically. It manages visual appearance and motion of a cell together. It is controlled only by time-sampling parameter. It is split into internally-induced stuff aka local interior affairs, externally-induced stuff aka global cell movement. Ulman & Svoboda (CBIA) ISBIq March 14, / 18

17 Design of the new simulator Single cell simulation Local interior affairs This stage supervises any changes that are relevant to the specific cell cycle phase, subject to cell current needs. Examples: re-organization of chromatin, drifts (if desired) or other visually apparent changes of cell nucleus or nucleoli, changes in cell shape, etc. Ulman & Svoboda (CBIA) ISBIq March 14, / 18

18 Design of the new simulator Single cell simulation Global cell movement This stage supervises any changes that are (usually) commonly happening during every cell cycle phase, (usually) influenced by cell environment. Examples: movement and shape changes of cell within its environment. Ulman & Svoboda (CBIA) ISBIq March 14, / 18

19 Design of the new simulator Cell population simulation The Scheduler::Run() Single cell life by means of iteratively executing specializations of the cycle phases template function. The Scheduler governs execution of these. Rule: The youngest cell is the first to be processed. Ulman & Svoboda (CBIA) ISBIq March 14, / 18

20 Techniques we used... Phases step by step G2-Phase Finish of cell growth, creation of centrosomes Brownian motion of dots Detection of major and minor axis inside cell shape using PCA Location of centrosomes driven by distance transform Universal motion of cell Ulman & Svoboda (CBIA) ISBIq March 14, / 18

21 Techniques we used... Phases step by step Prophase Chromatin condensates and nucleoli disappear Dots belonging to one chromosome tend to aggregate (move to mean position) and form highly condensed clusters Nucleoli masks are removed form the phantom Universal motion of cell Ulman & Svoboda (CBIA) ISBIq March 14, / 18

22 Techniques we used... Phases step by step Prophase Chromatin condensates and nucleoli disappear Dots belonging to one chromosome tend to aggregate (move to mean position) and form highly condensed clusters Nucleoli masks are removed form the phantom Universal motion of cell Ulman & Svoboda (CBIA) ISBIq March 14, / 18

23 Techniques we used... Phases step by step Metaphase Move of chromosomes to metaphase plate position 1 Nucleus membrane disappears 2 New positions for 23 chromosomes in metaphase plate are randomly generated 3 Old positions of 23 chromosomes, spread over the cell, are assigned to the new ones (min cost assignment problem Kuhn-Munkres algorithm) 4 The movement of chromosomes from old positions to the new ones is generated Ulman & Svoboda (CBIA) ISBIq March 14, / 18

24 Techniques we used... Phases step by step Metaphase Move of chromosomes to metaphase plate position 1 Nucleus membrane disappears 2 New positions for 23 chromosomes in metaphase plate are randomly generated 3 Old positions of 23 chromosomes, spread over the cell, are assigned to the new ones (min cost assignment problem Kuhn-Munkres algorithm) 4 The movement of chromosomes from old positions to the new ones is generated Ulman & Svoboda (CBIA) ISBIq March 14, / 18

25 Techniques we used... Phases step by step Metaphase Move of chromosomes to metaphase plate position 1 Nucleus membrane disappears 2 New positions for 23 chromosomes in metaphase plate are randomly generated 3 Old positions of 23 chromosomes, spread over the cell, are assigned to the new ones (min cost assignment problem Kuhn-Munkres algorithm) 4 The movement of chromosomes from old positions to the new ones is generated Ulman & Svoboda (CBIA) ISBIq March 14, / 18

26 Techniques we used... Phases step by step Metaphase Move of chromosomes to metaphase plate position 1 Nucleus membrane disappears 2 New positions for 23 chromosomes in metaphase plate are randomly generated 3 Old positions of 23 chromosomes, spread over the cell, are assigned to the new ones (min cost assignment problem Kuhn-Munkres algorithm) 4 The movement of chromosomes from old positions to the new ones is generated Ulman & Svoboda (CBIA) ISBIq March 14, / 18

27 Techniques we used... Phases step by step Anaphase Chromosomes are split and drawn to opposite poles Simple model of forces in which each chromosome is pulled to one centrosomes while the individual chromosomes push away each other In parallel the cell mask if slightly expanded along major axis using fast level set methods Universal motion of cell without additional deformations Ulman & Svoboda (CBIA) ISBIq March 14, / 18

28 Techniques we used... Phases step by step Anaphase Chromosomes are split and drawn to opposite poles Simple model of forces in which each chromosome is pulled to one centrosomes while the individual chromosomes push away each other In parallel the cell mask if slightly expanded along major axis using fast level set methods Universal motion of cell without additional deformations Ulman & Svoboda (CBIA) ISBIq March 14, / 18

29 Techniques we used... Phases step by step Anaphase Chromosomes are split and drawn to opposite poles Simple model of forces in which each chromosome is pulled to one centrosomes while the individual chromosomes push away each other In parallel the cell mask if slightly expanded along major axis using fast level set methods Universal motion of cell without additional deformations Ulman & Svoboda (CBIA) ISBIq March 14, / 18

30 Techniques we used... Phases step by step Telophase Creation of nucleus membrane, nucleoli; The chromatin is uncoiled Masks of two new daughter nuclei created Masks of new nucleoli in each nucleus created The chromatin is uncoiled via Brownian motion Universal motion of cell Ulman & Svoboda (CBIA) ISBIq March 14, / 18

31 Techniques we used... Phases step by step Telophase Creation of nucleus membrane, nucleoli; The chromatin is uncoiled Masks of two new daughter nuclei created Masks of new nucleoli in each nucleus created The chromatin is uncoiled via Brownian motion Universal motion of cell Ulman & Svoboda (CBIA) ISBIq March 14, / 18

32 Techniques we used... Phases step by step Telophase Creation of nucleus membrane, nucleoli; The chromatin is uncoiled Masks of two new daughter nuclei created Masks of new nucleoli in each nucleus created The chromatin is uncoiled via Brownian motion Universal motion of cell Ulman & Svoboda (CBIA) ISBIq March 14, / 18

33 Techniques we used... Phases step by step Telophase Creation of nucleus membrane, nucleoli; The chromatin is uncoiled Masks of two new daughter nuclei created Masks of new nucleoli in each nucleus created The chromatin is uncoiled via Brownian motion Universal motion of cell Ulman & Svoboda (CBIA) ISBIq March 14, / 18

34 Techniques we used... Phases step by step Cytokinesis Cell is pinched and new independent daughter cells are created PCA is used for detection for optimal cut (minor eigenvectors define cutting plane) Not yet completely implemented as this step is usually not noticeable (too short) Universal motion of cell without additional deformations Ulman & Svoboda (CBIA) ISBIq March 14, / 18

35 Techniques we used... Phases step by step G1-Phase Growth of new born daughter cell to full size An expanding flow field stretches all masks and moves dot positions Brownian motion of dots Universal motion of cell Ulman & Svoboda (CBIA) ISBIq March 14, / 18

36 Techniques we used... Phases step by step G1-Phase Growth of new born daughter cell to full size An expanding flow field stretches all masks and moves dot positions Brownian motion of dots Universal motion of cell Ulman & Svoboda (CBIA) ISBIq March 14, / 18

37 Techniques we used... Phases step by step G1-Phase Growth of new born daughter cell to full size An expanding flow field stretches all masks and moves dot positions Brownian motion of dots Universal motion of cell Ulman & Svoboda (CBIA) ISBIq March 14, / 18

38 Techniques we used... Phases step by step S-Phase Replication of DNA, i.e. growth of nucleus Brownian motion of dots Dots are duplicated per partes until all dots are duplicated Universal motion of cell Ulman & Svoboda (CBIA) ISBIq March 14, / 18

39 Techniques we used commonly Universal motion of cell It consists of rigid: translation and rotation non-rigid: decent local deformations of cell shape translation is preferred towards widest escape-tunnel in unconstrained environment: it simulates Brownian motion rotation is Gaussian with mean 0deg and sigma 13deg (+-15deg angles are with probability half of the probability for 0deg) inflating-deflating flow field that preserves volume in any case: coherency of motion is preserved Ulman & Svoboda (CBIA) ISBIq March 14, / 18

40 Techniques we used commonly Rendering of final images It consists of PSF simulation (simulates optics) Poissoned gained signal (uncertainty in the number of incoming photons) added little Poisson (dark current, params from docs) added Gaussian (read-out noise, params from docs) two sets of params: low (cca 0.5dB) and high (cca 0.8dB) SNR Ulman & Svoboda (CBIA) ISBIq March 14, / 18

41 Results demonstration of high and low SNR frames demonstration of phanthom video demonstration of final images video Ulman & Svoboda (CBIA) ISBIq March 14, / 18

42 References Simmons M. J., Snustad D. P. Genetika, Masarykova Univerzita, 2009 Reece J. B., Urry L. A., Cain M. L., Wasserman S. A., Minorsky P. V., Jackson R. B. Campbell Biology, 9th ed., Pearson Education, 2011 Krontorád-Koutná I., slides for course PV185 Biology Panorama I (autumn 2012) Liu L., Shell D. Assessing Optimal Assignment under Uncertainty: An Interval-based Algorithm. International Journal of Robotics Research (IJRR). vol. 30, no. 7, pp Jun 2011 Ulman & Svoboda (CBIA) ISBIq March 14, / 18

43 Any questions? Ulman & Svoboda (CBIA) ISBIq March 14, / 18