Academic
Publications
Automated Mitosis Detection of Stem Cell Populations in Phase-Contrast Microscopy Images

Automated Mitosis Detection of Stem Cell Populations in Phase-Contrast Microscopy Images,10.1109/TMI.2010.2089384,IEEE Transactions on Medical Imaging

Automated Mitosis Detection of Stem Cell Populations in Phase-Contrast Microscopy Images   (Citations: 3)
BibTex | RIS | RefWorks Download
Due to the enormous potential and impact that stem cells may have on regenerative medicine, there has been a rapidly growinginterestfortoolstoanalyzeandcharacterizethebehaviors of these cells in vitro in an automated and high throughput fashion. Among these behaviors, mitosis, or cell division, is important since stem cells proliferate and renew themselves through mitosis. How- ever, current automated systems for measuring cell proliferation often require destructive or sacrificial methods of cell manipula- tion such as cell lysis or in vitro staining. In this paper, we pro- pose an effective approach for automated mitosis detection using phase-contrast time-lapse microscopy, which is a nondestructive imaging modality, thereby allowing continuous monitoring of cells in culture. In our approach, we present a probabilistic model for event detection, which can simultaneously 1) identify spatio-tem- poral patch sequences that contain a mitotic event and 2) localize a birth event, defined as the time and location at which cell division is completed and two daughter cells are born. Our approach sig- nificantly outperforms previous approaches in terms of both detec- tion accuracy and computational efficiency, when applied to multi- potent C3H10T1/2 mesenchymal and C2C12 myoblastic stem cell populations.
Journal: IEEE Transactions on Medical Imaging - TMI , vol. 30, no. 3, pp. 586-596, 2011
Cumulative Annual
View Publication
The following links allow you to view full publications. These links are maintained by other sources not affiliated with Microsoft Academic Search.
    • ...To detect the birth events (time and location at which one cell divides into two cells), we have adopted the mitosis detection technique recently developed in [13]...

    Ryoma Biseet al. Reliable cell tracking by global data association

    • ...). Future work will incorporate additional algorithms that detect and measure cell behaviors such as mitosis ...

    Dai Fei Elmer Keret al. An Engineered Approach to Stem Cell Culture: Automating the Decision P...

    • ...Other approaches do not involve cell tracking [3, 11, 12, 7]. Most of these methods first reduce search space by detecting candidates that are likely to contain mitosis [3, 12, 7]. Probabilistic or statistical models are often employed to identify true mitosis among candidates by learning visual change of mitotic cells based on human-annotated samples [11, 12, 7]...
    • ...Other approaches do not involve cell tracking [3, 11, 12, 7]. Most of these methods first reduce search space by detecting candidates that are likely to contain mitosis [3, 12, 7]. Probabilistic or statistical models are often employed to identify true mitosis among candidates by learning visual change of mitotic cells based on human-annotated samples [11, 12, 7]...
    • ...Other approaches do not involve cell tracking [3, 11, 12, 7]. Most of these methods first reduce search space by detecting candidates that are likely to contain mitosis [3, 12, 7]. Probabilistic or statistical models are often employed to identify true mitosis among candidates by learning visual change of mitotic cells based on human-annotated samples [11, 12, 7]...
    • ...Recently, one mitosis detection approach demonstrated success in precisely detecting birth events that are defined as the time and location at which mitosis is complete and two daughter cells first appear [7]...
    • ...Before these steps, we apply a preprocessing method [7] to correct intrinsic illumination variation in phase-contrast microscopy images...
    • ...Unique scale histograms [7] are used as visual features...
    • ...Note that previous patch sequence construction methods [3, 12, 7] do not have such capability; as a result, their Figure 3. An example of a candidate patch sequence...
    • ...Conditional Random Fields (HCRF) [17] and Event Detection Conditional Random Fields (EDCRF) [7], both of which have been used for mitosis detection...
    • ...Following previous works [17, 7], state and transition functions are defined as...
    • ...We use unique scale gradient histograms [7] as visual features in our experiments...
    • ...Threshold th=1 th=3 Approach [7] Our approach [7] Our approach...
    • ...Threshold th=1 th=3 Approach [7] Our approach [7] Our approach...
    • ...Table 1. Mitosis detection performance comparison between the previous approach [7] and the proposed approach on C2C12 and BAEC in terms of F-measure and AUC of the PR-curve...
    • ...We compare our approach with the previous work [7], which constructs candidate patch sequences by thresholding brightness without the candidate birth event detection step...
    • ...To the best of our knowledge, [7] is the only work that explicitly detects birth events during mitosis...
    • ...We also compare our TL-HCRF model with HCRF [18] and ED-CRF [7] on the task that classifies candidate patch sequences whether each of them contains a birth event or not...
    • ...Our approach significantly outperforms the previous approach [7] in terms of F-measure and AUC of the PR graph as shown in Table 1. In particular, performance improvement is significant on BAEC which is at 100% confluence from the beginning to the end...
    • ...To make the inference of all sub-labels tractable, EDCRF restricts that each sub-class label is associated only with hidden states in a disjoint set [7], which may degrade the overall performance...

    Seungil Huhet al. Detection of mitosis within a stem cell population of high cell conflu...

Sort by: